Strategies to improve smoking cessation rates in primary care

Nicola Lindson; Gillian Pritchard; Bosun Hong; Thomas R Fanshawe; Andrew Pipe; Sophia Papadakis

doi:10.1002/14651858.CD011556.pub2

. 2021 Sep 6;2021(9):CD011556. doi: 10.1002/14651858.CD011556.pub2

Strategies to improve smoking cessation rates in primary care

Nicola Lindson ¹, Gillian Pritchard ^2,³, Bosun Hong ⁴, Thomas R Fanshawe ¹, Andrew Pipe ², Sophia Papadakis ^2,^✉

Editor: Cochrane Tobacco Addiction Group

PMCID: PMC8543670 PMID: 34693994

Abstract

Background

Primary care is an important setting in which to treat tobacco addiction. However, the rates at which providers address smoking cessation and the success of that support vary. Strategies can be implemented to improve and increase the delivery of smoking cessation support (e.g. through provider training), and to increase the amount and breadth of support given to people who smoke (e.g. through additional counseling or tailored printed materials).

Objectives

To assess the effectiveness of strategies intended to increase the success of smoking cessation interventions in primary care settings.

To assess whether any effect that these interventions have on smoking cessation may be due to increased implementation by healthcare providers.

Search methods

We searched the Cochrane Tobacco Addiction Group's Specialized Register, the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase, and trial registries to 10 September 2020.

Selection criteria

We included randomized controlled trials (RCTs) and cluster‐RCTs (cRCTs) carried out in primary care, including non‐pregnant adults. Studies investigated a strategy or strategies to improve the implementation or success of smoking cessation treatment in primary care. These strategies could include interventions designed to increase or enhance the quality of existing support, or smoking cessation interventions offered in addition to standard care (adjunctive interventions). Intervention strategies had to be tested in addition to and in comparison with standard care, or in addition to other active intervention strategies if the effect of an individual strategy could be isolated. Standard care typically incorporates physician‐delivered brief behavioral support, and an offer of smoking cessation medication, but differs across studies. Studies had to measure smoking abstinence at six months' follow‐up or longer.

Data collection and analysis

We followed standard Cochrane methods. Our primary outcome ‐ smoking abstinence ‐ was measured using the most rigorous intention‐to‐treat definition available. We also extracted outcome data for quit attempts, and the following markers of healthcare provider performance: asking about smoking status; advising on cessation; assessment of participant readiness to quit; assisting with cessation; arranging follow‐up for smoking participants. Where more than one study investigated the same strategy or set of strategies, and measured the same outcome, we conducted meta‐analyses using Mantel‐Haenszel random‐effects methods to generate pooled risk ratios (RRs) and 95% confidence intervals (CIs).

Main results

We included 81 RCTs and cRCTs, involving 112,159 participants. Fourteen were rated at low risk of bias, 44 at high risk, and the remainder at unclear risk.

We identified moderate‐certainty evidence, limited by inconsistency, that the provision of adjunctive counseling by a health professional other than the physician (RR 1.31, 95% CI 1.10 to 1.55; I² = 44%; 22 studies, 18,150 participants), and provision of cost‐free medications (RR 1.36, 95% CI 1.05 to 1.76; I² = 63%; 10 studies,7560 participants) increased smoking quit rates in primary care. There was also moderate‐certainty evidence, limited by risk of bias, that the addition of tailored print materials to standard smoking cessation treatment increased the number of people who had successfully stopped smoking at six months' follow‐up or more (RR 1.29, 95% CI 1.04 to 1.59; I² = 37%; 6 studies, 15,978 participants).

There was no clear evidence that providing participants who smoked with biomedical risk feedback increased their likelihood of quitting (RR 1.07, 95% CI 0.81 to 1.41; I² = 40%; 7 studies, 3491 participants), or that provider smoking cessation training (RR 1.10, 95% CI 0.85 to 1.41; I² = 66%; 7 studies, 13,685 participants) or provider incentives (RR 1.14, 95% CI 0.97 to 1.34; I² = 0%; 2 studies, 2454 participants) increased smoking abstinence rates. However, in assessing the former two strategies we judged the evidence to be of low certainty and in assessing the latter strategies it was of very low certainty. We downgraded the evidence due to imprecision, inconsistency and risk of bias across these comparisons. There was some indication that provider training increased the delivery of smoking cessation support, along with the provision of adjunctive counseling and cost‐free medications. However, our secondary outcomes were not measured consistently, and in many cases analyses were subject to substantial statistical heterogeneity, imprecision, or both, making it difficult to draw conclusions.

Thirty‐four studies investigated multicomponent interventions to improve smoking cessation rates. There was substantial variation in the combinations of strategies tested, and the resulting individual study effect estimates, precluding meta‐analyses in most cases. Meta‐analyses provided some evidence that adjunctive counseling combined with either cost‐free medications or provider training enhanced quit rates when compared with standard care alone. However, analyses were limited by small numbers of events, high statistical heterogeneity, and studies at high risk of bias. Analyses looking at the effects of combining provider training with flow sheets to aid physician decision‐making, and with outreach facilitation, found no clear evidence that these combinations increased quit rates; however, analyses were limited by imprecision, and there was some indication that these approaches did improve some forms of provider implementation.

Authors' conclusions

There is moderate‐certainty evidence that providing adjunctive counseling by an allied health professional, cost‐free smoking cessation medications, and tailored printed materials as part of smoking cessation support in primary care can increase the number of people who achieve smoking cessation. There is no clear evidence that providing participants with biomedical risk feedback, or primary care providers with training or incentives to provide smoking cessation support enhance quit rates. However, we rated this evidence as of low or very low certainty, and so conclusions are likely to change as further evidence becomes available. Most of the studies in this review evaluated smoking cessation interventions that had already been extensively tested in the general population. Further studies should assess strategies designed to optimize the delivery of those interventions already known to be effective within the primary care setting. Such studies should be cluster‐randomized to account for the implications of implementation in this particular setting. Due to substantial variation between studies in this review, identifying optimal characteristics of multicomponent interventions to improve the delivery of smoking cessation treatment was challenging. Future research could use component network meta‐analysis to investigate this further.

Plain language summary

Are there ways to improve stop‐smoking treatment in primary care to help more people to quit smoking?

What is stop‐smoking treatment in primary care?

Primary care, also known as family medicine or general practice, is where people go to see a health professional for mostly day‐to‐day health issues. It is one of the best places for people who smoke tobacco to get help to quit. When people visit primary care they may be asked if they smoke. If they do, they may then be helped to quit, typically through counseling and medications.

Why we did this Cochrane Review

Support to stop smoking in primary care is not always delivered well or consistently. Health providers may be unsure how best to deliver treatment, may have limited time to deliver it, or lack the resources needed. Ways to improve the delivery and success of stop‐smoking support in primary care have been suggested. Some of these are designed to make sure the treatment already available is delivered often and well, e.g. training providers on how best to help people quit, and some are designed to increase the support available for participants, e.g. providing additional counseling and printed materials. Our aim was to look at which of these approaches works best on their own or together.

What did we do?

We searched for studies that looked at ways to improve standard stop‐smoking support within primary care, and where the treatments people received were decided at random.

We wanted to find out:

‐ how many people were asked about their smoking and provided with advice and support;

‐ how many people tried to quit smoking; and

‐ how many people stopped smoking for at least six months.

We included evidence published to 10th September 2020.

What we found

We found 81 studies including 112,159 smokers in primary care patients. Studies looked at many ways to improve the delivery and success of stop‐smoking support in primary care. Some looked at just one strategy, and some looked at two or more in combination. More than one study looked at each of the following individual strategies: additional counseling; free medications; feedback to participants on markers of their individual health risk linked to smoking; printed materials tailored to participants; health provider training; and rewards to health providers for providing support.

Most studies took place in Europe (39 studies) and the USA (26 studies).

What are the results of our review?

More people probably stop smoking for at least six months when they are given additional counseling (22 studies, 18,150 people), free stop‐smoking medications (10 studies, 7560 people), or printed materials tailored to them (6 studies, 15,978 people), as part of stop‐smoking support in primary care. We are uncertain whether providing people with feedback on markers of their individual health risk, providing healthcare providers with training, or with rewards for providing stop‐smoking support, help more people to quit.

Thirty‐four studies looked at more than one strategy to improve stop‐smoking treatment in primary care. Combinations differed greatly across studies, with different levels of success, and it was not possible to draw conclusions on what worked best.

There was not enough information to help us clearly understand whether there were increases in the amount of stop‐smoking support provided or increases in the numbers of people making a quit attempt.

How reliable are these results?

For some of our results the data varied widely, for some there was not enough data, and in some cases there were quality issues with included studies.

We are moderately confident that people are more likely to quit smoking if someone in addition to the primary care doctor also provides stop‐smoking counseling, if free stop‐smoking medications are provided, or if printed materials tailored to the participant are provided as part of stop‐smoking support offered in primary care. However, results might change as further evidence becomes available.

We are less confident about the effectiveness of providing people with feedback on markers of their individual health risk, giving healthcare providers training on stop‐smoking treatments, or giving healthcare providers rewards for giving stop‐smoking support. These results are likely to change when more evidence becomes available.

Summary of findings

Background

Description of the condition

Tobacco use is the leading cause of premature death worldwide (World Health Organization 2017). From a chronic illness perspective, people who smoke have a 50% to 70% greater chance of dying from stroke or coronary heart disease than people who do not, and 85% of cancers of the trachea, bronchus, and lung are directly attributable to tobacco use (U.S. Department of Health and Human Services 2014). Tobacco use is also a leading risk factor for other major causes of death, including 16 types of cancer, chronic obstructive pulmonary disease, and lower respiratory tract infections (U.S. Department of Health and Human Services 2014; World Health Organization 2017).

There is overwhelming evidence to support both the health and economic benefits of smoking cessation. If a person smokes, supporting them with quitting is the single most effective intervention a clinician can provide to reduce the risk of premature disease, disability and death (Fiore 2008; Royal College of Physicians 2018; Tengs 1995). Quitting smoking reduces the excess risk of smoking‐related coronary heart disease by approximately 50% within one year, and to normal levels within five years (U.S. Department of Health and Human Services 2014). Smoking cessation is also considered to be among the most cost‐effective preventive interventions available to clinicians and health systems (Tengs 1995; Cromwell 1997; Roncker 2005; Franco 2007; Gaziano 2007; Royal College of Physicians 2018; U.S. Department of Health and Human Services 2020).

Description of the intervention

Primary care practice, also known as family medicine or general practice, has been identified as an important setting for intervening with tobacco users because of the large reach of primary care settings, the long‐term relationships with patients and their role in addressing disease prevention (U.S. Department of Health and Human Services 2020; World Health Organization 2020).

Evidence‐based guidelines for the delivery of tobacco treatment emphasize the important role of primary care clinicians in tobacco treatment delivery (Verbiest 2017). The World Health Organization (WHO) and other international authorities have called for smoking cessation to be integrated into primary health care globally, as it is seen as the most suitable health system 'environment' for providing advice and support on smoking cessation (Fiore 2008; World Health Organization 2008; World Health Organization 2020; U.S. Department of Health and Human Services 2020). Specifically, the combination of behavioral support and stop‐smoking pharmacotherapy have been shown to significantly enhance long‐term cessation rates; it follows that increasing the use of these evidence‐based treatments is an important target (Stead 2016). While models of delivery differ across international settings, clinical practice guidelines recommend that primary care providers support people who smoke with quitting by: asking them about their smoking status, providing advice on quitting to those identified as smoking, and supporting cessation by offering behavioral counseling and/or pharmaceutical treatment or both when smokers identify themselves as ready to quit (Verbiest 2017). See Secondary outcomes below for more information.

However, there is a well‐documented ‘practice gap’ in the rates at which smoking cessation is addressed by practitioners in clinical settings. International studies have documented that between 40% and 70% of people who smoke report having received cessation advice from their physician (Bartsch 2016; Papadakis 2014; Reid 2019; World Health Organization 2020). While practitioners tend to deliver advice to quit at moderate rates, studies have shown that the rates of providing specific assistance (i.e. behavioral counseling, printed self‐help materials, stop‐smoking medications, or follow‐up support) are much lower (Bartsch 2016; Papadakis 2014). When it is offered, the amount and breadth of assistance is also likely to differ considerably across practices, which may have an effect on rates of smoking cessation.

How the intervention might work

Several barriers to optimal cessation practice in primary care have been identified at the patient, provider, and practice levels (Martin‐Cantera 2020; Van Rossem 2015; Vogt 2005; Young 2001). Identified barriers include a lack of knowledge and skills among providers, provider attitudes and perceptions, lack of time and organizational supports, and a lack of patient motivation and other patient‐level factors. Interventions which address these barriers are expected to enhance rates of tobacco treatment delivery by primary care providers, increase the use of evidence‐based stop smoking treatment by patients, and subsequently lead to enhanced quit rates among patients identified in primary care (Van Rossem 2015; Martin‐Cantera 2020; Vogt 2005; Young 2001).

Strategies to improve the delivery of standard smoking cessation support in primary care could include the provision of provider training, real‐time counseling prompts, and provider performance feedback. These examples represent strategies that span practice‐ and provider‐implementation levels. Another way to boost smoking quit rates in primary care could be to incorporate additional intervention components alongside those already commonly delivered as part of standard care, e.g. provision of tailored print materials, adjunctive counseling provided by allied health professionals and providing people with specific feedback about their smoking‐related health risks. These strategies could be implemented either individually or as part of a multicomponent intervention (combining more than one strategy). While there is a lack of implementation knowledge to inform the design and delivery of tobacco treatment interventions in primary care practice, multicomponent interventions have previously been shown to be the most effective method for increasing both provider performance in the delivery of smoking cessation treatment and improving cessation rates among participants (Anderson 2004; Fiore 2008; Grimshaw 2001; Martin‐Cantera 2015; Papadakis 2010). They are designed to address several barriers to treatment delivery in a synergistic manner, acknowledging the need for more complex or sophisticated intervention models, or both, to bring about changes in healthcare practice and behavior.

Why it is important to do this review

Reflecting the challenges surrounding the effective implementation of smoking cessation treatment in primary care, much research has been carried out investigating how to improve both the implementation and success of these interventions. Some have focused on practice‐level interventions (such as electronic medical record prompts or outreach facilitation (Cummings 1989a; Verbiest 2014); some have focused on provider‐level interventions, such as provider training and incentives (Lennox 1998; Olano Espinosa 2013; Roski 2003), and some have focused on patient‐level interventions (over and above the standard advice delivered by primary care physicians; such as adjunctive counseling, cost‐free medications, biomedical feedback, and tailored printed materials; An 2006; Meyer 2008; Minué‐Lorenzo 2019; Ronaldson 2018). Others have tested a combination of these approaches in multicomponent interventions (e.g. Katz 2004; Twardella 2007; Unrod 2007). Bringing this evidence together allows us to summarize the research methodologies used and to synthesize the evidence in support of specific strategies, or the combination of strategies, that are effective in increasing rates of smoking cessation in the primary care setting. This can be used to inform both clinical practice and the implementation of health policy. Several published meta‐analyses have examined the effect of physician advice and other provider interventions on smoking cessation, but many of these reviews have not been specific to the primary care setting (Boyle 2014; Carson 2012; Clair 2019; Fiore 2008; Rice 2017; Stead 2013; Van den Brand 2017). These previous reviews have also focused on the effect of providing advice on smoking abstinence only; they have not examined improvements in provider performance in the delivery of evidence‐based smoking cessation treatments that may have ultimately led to any increase in effectiveness. Two published meta‐analyses have aimed to do this: Anderson 2004 reviewed the literature published up to 2001, and Papadakis 2010 published an update which examined the literature prior to 2009. Additionally, Martin‐Cantera 2015 narratively reviewed the literature examining multicomponent interventions in primary care, published up to 2014. This review provides an up‐to‐date synthesis of the literature in this field.

Objectives

To assess the effectiveness of strategies intended to increase the success of smoking cessation interventions in primary care settings.

To assess whether any effect that these interventions have on smoking cessation may be due to increased implementation by healthcare providers.

Methods

Criteria for considering studies for this review

Types of studies

Randomized controlled trials (RCTs), cluster‐RCTs (cRCTs).

Types of participants

Participants include adult primary healthcare patients. For the purposes of this review, we defined primary care as family medicine or general medical practice. We did not include public health or community interventions in our definition of primary care, nor did we assess interventions delivered in dental offices or pharmacies. We included trials which covered the whole practice population, as well as those which included specific subpopulations recruited from primary care settings (e.g. people with chronic obstructive pulmonary disease (COPD), or people with diabetes). We did not include studies that solely addressed the behavior of pregnant women or adolescents, as they are addressed by other Cochrane Reviews (Chamberlain 2017; Claire 2020; Fanshawe 2017).

For our primary outcome, and most of our secondary outcomes, all participants were required to be people who used tobacco at study baseline. However, for our secondary outcome, 'Number of patients asked whether they smoke', participants could include the general population of primary care patients (i.e. both people who used tobacco and people who did not use tobacco at baseline).

Types of interventions

To be included in this review, studies must have investigated an intervention strategy or strategies designed to improve the implementation or success of smoking cessation treatment in primary care. The interventions under investigation in this review were therefore not standard smoking cessation support incorporating brief advice delivered by a primary care physician, or the standard provision of smoking cessation medications in primary care. Interventions of interest could include any strategy or strategies designed to increase or enhance the quality of the support offered, or an adjunctive smoking cessation intervention offered in addition to standard care. Interventions could be implemented at any level (i.e. practice, provider or participant) and the patient‐level components could be delivered by any health professional within a primary care practice setting. Examples of patient‐level interventions investigated in this review included adjunctive counseling delivered by a health professional other than the physician, cost‐free smoking cessation medications and the provision of tailored print materials. Examples of provider‐level interventions included provider training and provider incentives. Examples of practice‐level interventions were outreach facilitation and electronic medical record (EMR) prompts. The categorization of these interventions is subjective, and some interventions may fit equally well at more than one level. For example, we considered in detail whether cost‐free medications should be categorized as a patient‐level, practice‐level, or system‐level intervention (e.g. where medication costs are government‐subsidized), and decided that it could be categorized as all of these. We decided on patient‐level in this instance, as the participant is the beneficiary of the cost savings, which have the potential to increase medication use.

Valid intervention groups were tested as an adjunct to and in comparison with 'standard' smoking cessation support or 'usual care', in order to test the effect of the additional implementation strategy over and above standard care. Standard care is defined differently within and across different communities and studies; however, it typically involves brief behavioral support from the primary care physician, alongside an offer of smoking cessation medication. We also included studies with head‐to‐head comparisons of two or more active interventions, but only if it was possible to isolate the effects of a single strategy or component designed to enhance the delivery of tobacco cessation treatment in primary care.

We did not include studies which covered interventions to enhance tobacco treatment delivery as part of a multifactorial lifestyle intervention.

Types of outcome measures

Primary outcomes

The primary outcome measure was smoking abstinence at long‐term follow‐up in participants who reported smoking at baseline. To be eligible for inclusion, studies had to measure smoking status at least six months from the start of the intervention. We excluded studies with abstinence measured at less than six months' follow‐up.

In trials with more than one measure of abstinence, we preferred the measure using the longest follow‐up and the strictest criteria, in line with the Russell Standard (West 2005). We used sustained or continuous abstinence over point prevalence abstinence, and biochemically‐validated abstinence, such as exhaled carbon monoxide (CO), over self‐report. We favored biochemically‐validated point prevalence abstinence over self‐reported continuous or prolonged abstinence. We considered participants lost to follow‐up to be still smoking, in line with the practice of the Cochrane Tobacco Addiction Group.

We chose smoking abstinence as the primary outcome, as this is the most clinically relevant outcome; an increase in the number of people quitting is the ultimate goal of any attempt to increase the implementation of smoking cessation treatment.

Secondary outcomes

Our secondary outcomes are deemed to be process outcomes. We therefore did not include studies that only reported on our secondary outcomes and did not investigate our primary outcome.

Provider performance in tobacco treatment delivery (these outcomes were informed by the 5As; a sequence of actions proposed by US smoking cessation guidelines that can be applied in primary care settings; Fiore 2008)
- Number of participants asked whether they smoke (the denominator for this also includes participants who were not smoking at baseline in studies that enrolled people who smoked and people that did not);
- Number of participants identified as smoking who were advised to quit;
- Number of participants identified as smoking whose readiness to quit was assessed;
- Number of participants identified as smoking who were assisted to quit (further divided into general assistance, medications prescribed, quit date set, counseling provided, self‐help materials provided);
- Number of participants identified as smoking who had follow‐up appointments arranged to address smoking.
Participant quit attempts, as defined by individual studies.

Search methods for identification of studies

Electronic searches

We searched the following databases to 10th September 2020:

Cochrane Tobacco Addiction Group Specialized Register;
Cochrane Central Register of Controlled Trials (CENTRAL);
MEDLINE (via PubMed);
Embase.

The search strategy used the following keyword terms: ('smoking' or 'smoking cessation' or 'tobacco‐use cessation', or 'tobacco‐use‐disorder) AND ('primary health care' or 'physicians' or 'family practice' or 'general practice' or 'general practitioners' or 'physicians, family'). We used standard search strings, using the Cochrane Highly Sensitive Search Strategy for identifying randomized controlled trials, as well as 'controlled trials' or 'evaluation studies'. We applied no restrictions by language or publication status. See Appendix 1 and Appendix 2 for the example PubMed and Specialized Register search strategies respectively.

Searching other resources

We searched the following trial registers: www.clinicaltrials.gov and the International Clinical Trials Registration Platform (WHO ICTRP), and reference lists of eligible studies. We also contacted study authors for unpublished results of completed studies.

Data collection and analysis

Selection of studies

SP, GP and BH independently reviewed titles and abstracts of reports for possible inclusion. We reviewed the full text of any reports which could not be fully assessed using the title and abstract, along with any reports that appeared to be eligible based on the available information. Two review authors (from SP, GP and BH) then independently assessed all of the full‐text articles retrieved, and resolved discrepancies by discussion with a third review author (AP or NL), who acted as an arbiter. We then linked multiple reports of the same eligible study. We recorded all reports of studies excluded at the full‐text screening phase, together with the reason for exclusion.

Data extraction and management

One review author (from SP, GP, BH) extracted data on study characteristics of eligible studies. Two review authors (from SP, GP, BH) independently extracted data on outcomes, and categorized studies according to the type and level of intervention. We extracted the following information from each of the included studies:

lead author and year of publication;
country in which intervention was delivered;
methods of recruiting healthcare practices and participants within practices;
inclusion criteria, including subpopulations;
type of study design (RCT, cluster‐RCT);
target of intervention (participant, provider, practice);
data collection method (interview, telephone, mail survey);
characteristics of study participants (age, sex, comorbidities, readiness to quit);
duration of intervention (in weeks);
details of the intervention;
description of the comparator intervention;
outcomes measured, including definitions used and time point at which they were assessed (in weeks);
use of biochemical validation and participant response rate;
methods used to manage missing data;
for each outcome: number of participants in each arm; loss to follow‐up rate; number of events in each arm; intra‐class correlation coefficient (ICC) (cluster‐RCTs only);
study funding source;
authors' declarations of interest.

Methods for categorizing details of intervention

We categorized intervention strategies into three groups, based on the level at which they were designed to intervene (i.e. participant, provider, practice). We further categorized interventions as either a single or a multicomponent intervention. For the purposes of this review, we defined single‐component interventions as those which included only one intervention strategy. We defined multicomponent interventions as interventions which included two or more intervention strategies, at any level. We used a preliminary list of intervention strategies based on previous systematic reviews (Anderson 2004; Fiore 2008; Papadakis 2010) with further categories added as appropriate to describe other intervention modalities identified in the literature.

Assessment of risk of bias in included studies

Two review authors (from SP, GP, BH, NL) independently assessed the risk of bias of the included studies, using Cochrane's RoB1 (Higgins 2011).

We assessed the following domains:

sequence generation (as an indicator of selection bias)
allocation concealment (as an indicator of selection bias)
blinding of outcome assessors (as an indicator of detection bias)
incomplete outcome data (as an indicator of attrition bias)

We did not assess any indicators of performance bias, as all of the studies were assessing a behavioral strategy, and therefore it would have been impossible to blind research staff and participants.

We also assessed the following other sources of bias for c‐RCTs only:

recruitment bias due to recruitment of participants to clusters after allocation;
unbalanced baseline characteristics;
whether statistical adjustment had been made to the analysis to account for the potential correlation of effects within clusters.

Measures of treatment effect

For each study and outcome (smoking abstinence, physician performance outcomes, quit attempts) we calculated the risk ratio (RR) and 95% confidence interval (CI) for each relevant comparison investigated (intervention group versus control group). For smoking abstinence and quit attempts the denominators were the number of people randomized to each study arm, assuming that any participants lost to follow‐up were continuing to smoke or had not made a quit attempt. For the physician performance outcomes we carried out a complete case analysis where possible.

Unit of analysis issues

All analyses contain participant‐level data from both RCTs and c‐RCTs. We investigated the effect of adjusting for clustering in c‐RCTs by inflating the standard error of the log RR using the design effect calculated from the estimated ICC that was reported in the study. If no ICC was reported, we assumed a typical ICC value for smoking cessation trials, based on Baskerville 2001. See below (Sensitivity analysis) for more details.

Dealing with missing data

We recorded the proportions of participants lost to follow‐up in each relevant arm of included studies and used this information in our risk of bias assessments. Any participants with missing smoking or quit attempts data at follow‐up were deemed to have returned to active smoking or to have not made a quit attempt respectively, and were included in the denominator for calculating the risk ratio. We did not impute missing data for physician performance outcomes.

Assessment of heterogeneity

We assessed statistical heterogeneity within meta‐analyses and between subgroups using the I² statistic (Higgins 2003). We considered an I² value greater than 50% to indicate moderate to substantial heterogeneity. Where an I² of greater than 75% was recorded for the pooled result of a meta‐analysis, and this remained unexplained by subgroup analyses, we judged whether it was appropriate to present the pooled estimate.

Assessment of reporting biases

Where analyses included 10 or more studies we generated funnel plots to investigate potential publication bias.

Data synthesis

We grouped studies by intervention type. Where there was more than one study testing an intervention type(s), and where appropriate, we performed meta‐analyses using Mantel‐Haenszel random‐effect models for each outcome. In studies that tested a single intervention component, we did not calculate a pooled estimate across intervention types for each level of intervention (e.g. participant, provider, practice) due to clinical heterogeneity. We were able to carry out meta‐analyses of our primary outcome for comparisons investigating the following singular intervention types:

Participant‐level:

adjunctive counseling
cost‐free medications
biomedical feedback (e.g. spirometry, CO monitoring)
tailored print materials

Provider‐level:

provider training
provider incentives

Some of the studies included in the analyses tested the intervention components above alongside standard care and also used standard care as the comparator, whereas other studies tested the intervention component as part of a multicomponent intervention with a comparator that received the same multicomponent intervention minus the intervention component of interest. These two types of studies were combined in the same meta‐analysis with subgrouping to investigate whether the study design had an impact on study findings.

We were also able to meta‐analyze the following multicomponent interventions versus standard care alone, where more than one study tested the same combination of components:

adjunctive counseling and cost‐free medications
adjunctive counseling and provider training
provider training and flow sheet
provider training and outreach facilitation

Again we carried out meta‐analyses using random‐effects Mantel‐Haenszel methods to calculate RRs and 95% CIs. Where it was not possible to conduct meta‐analyses, i.e. there was only one study investigating the comparison, we summarized studies narratively, calculating and presenting their individual RRs and 95% CIs.

Subgroup analysis and investigation of heterogeneity

We used subgroup analyses to investigate any differences in the effects observed between studies for the primary smoking abstinence outcome, where possible:

for all comparisons: studies that tested the intervention component(s) of interest alongside and in comparison with standard care only versus studies that tested the intervention component(s) of interest alongside and in comparison with other intervention component(s) of interest;
for all comparisons: participants with chronic disease versus those without chronic disease, e.g. diabetes, COPD;
for adjunctive counseling comparisons: provider type; intensity; mode;
for biomedical feedback: type of biomedical feedback, e.g. spirometry, CO monitoring;
for tailored print materials: theoretical basis of tailoring.

Sensitivity analysis

We used sensitivity analyses to examine the effects of excluding studies with the following characteristics for the primary smoking abstinence outcome:

studies deemed to be at high risk of bias (i.e. judged to be at high risk for at least one risk of bias domain);
individually randomized studies (as opposed to c‐RCTs). c‐RCTs provide the best evidence for the effects of the interventions tested when implemented in primary care, as they would be in the 'real world'.

We also carried out two sensitivity analyses adjusting for appropriate estimates of ICCs in those c‐RCTs that did not report controlling for the clustered nature of the design, or those in which the ICC was not reported. Separate sensitivity analyses used estimated ICCs of 0.01 and 0.05 respectively (Baskerville 2001).

Summary of findings and assessment of the certainty of the evidence

We used GRADEpro GDT to import data from Review Manager 5 in order to create summary of findings tables (GRADEpro GDT; Review Manager 2020) for comparisons investigating the following single intervention components:

adjunctive counseling;
cost‐free medications;
biomedical feedback;
tailored print materials;
provider training;
provider incentives.

A summary of the intervention effect for the primary smoking abstinence outcome was produced for each comparison, and we used the GRADE approach to assess the certainty of the body of evidence (Schünemann 2013), as recommended in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2021). We used the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness, and publication bias) to assess the certainty of the body of evidence for our primary outcome ‐ smoking abstinence, downgrading by one level for serious or by two levels for very serious limitations for each consideration.

Results

Description of studies

Results of the search

Our searches identified 4518 non‐duplicate records. We screened all records and retrieved the full‐text papers of 824 potentially relevant articles. After screening the full texts we included 81 studies (see Characteristics of included studies), and identified 18 ongoing studies (see Characteristics of ongoing studies) and four studies awaiting classification (Studies awaiting classification). Figure 1 presents the PRISMA flow chart for this review.

Included studies

Of the 81 included studies 43 were individually‐randomized RCTs, 37 were c‐RCTs, and one was a factorial trial. Thirty‐nine studies were conducted in Europe, 26 in the USA, seven in Australia, two each in South America, South Korea and Canada, and one each in China, Pakistan and Thailand. All studies were conducted in primary care settings.

Participants

The 81 included studies represented 112,159 participants. Individual study sample sizes ranged from 48 to 6856. For all but one of the outcomes, all participants were people who attended the practices as patients and smoked tobacco at study baseline. However, where studies measured the number of patients who were asked whether they smoked, participants contributing to the outcome included anyone attending the primary care practice. These non‐smoking participants are included in the total number of participants specified above, but very few studies assessed this outcome.

Five studies recruited from specific population groups; two of these specifically recruited participants with COPD, one recruited participants with diabetes and hypertension, another participants with diabetes only, and the final study recruited participants with a low or moderate household income. The average age of participants ranged from 33 to 64 across studies, and average cigarettes per day ranged from 14 to 26.

Interventions and comparators

We classified study arms according to whether they offered any interventions designed to improve the delivery or success of smoking cessation treatment, over and above standard care. Standard smoking cessation support typically involves brief advice from a physician with a potential offer of medication, and generic printed self‐help materials. Some study arms offered multiple additional components (multicomponent interventions), whereas others offered a single additional component. We classified these components as either patient‐level, provider‐level or practice‐level. Within these classifications we identified the intervention types listed in the table below (for more detailed definitions of the strategies listed see Appendix 3):

PATIENT‐LEVEL	PROVIDER‐LEVEL	PRACTICE‐LEVEL
Adjunctive counseling (offered by a health professional other than the primary care physician, i.e. via a practice nurse, counselor, or smoking quitline)	Provider training	Modified vital sign stamp
Tailored printed materials	Provider performance audit and feedback	Treatment flow sheets/consult forms
Biomedical feedback (including CO monitoring, gene testing for lung cancer, spirometry and a combination of CO monitoring and spirometry)	Provider incentives	Electronic medical record (EMR) and decision support
Medication prompts	‐	Outreach facilitation
Patient incentives	‐	‐
SMS and Internet cessation programs	‐	‐
Information videos	‐	‐
Access to cost‐free medications (as opposed to medications with a fee, which would be considered standard care)	‐	‐
Proactive outreach	‐	‐

Adjunctive counseling in addition to standard smoking cessation care in primary care
Patient or population: people who attend primary care and smoke tobacco Setting: primary care (Australia, Europe, South Korea, United States) Intervention: adjunctive counseling plus standard or multicomponent smoking cessation support Comparison: standard or multicomponent smoking cessation support
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with control	Risk with adjunctive counseling
Smoking abstinence at 6‐month follow‐up or more. All studies	Study population		RR 1.31 (1.10 to 1.55)	18,150 (22 RCTs)	⊕⊕⊕⊝ MODERATE^a	‐
	7 per 100	9 per 100 (8 to 11)
Smoking abstinence at 6‐month follow‐up or more. Subgroup comparator: standard care	Study population		RR 1.43 (1.15 to 1.78)	12,852 (17 RCTs)	⊕⊕⊕⊝ MODERATE^b	‐
	4 per 100	6 per 100 (5 to 8)
Smoking abstinence at 6‐month follow‐up or more. Subgroup comparator: multicomponent intervention	Study population		RR 1.04 (0.87 to 1.23)	5298 (5 RCTs)	⊕⊕⊝⊝ LOW^c	‐
	14 per 100	14 per 100 (12 to 17)
*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).  CI: Confidence interval; RCT: Randomized controlled trial; RR: Risk ratio
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

Cost‐free medications used in addition to standard care in primary care
Patient or population: people who attend primary care and smoke tobacco Setting: primary care (Australia, Europe, Pakistan, United States) Intervention: cost‐free medications plus standard or multicomponent smoking cessation support Comparison: standard or multicomponent smoking cessation support
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with placebo	Risk with cost‐free medications
Smoking abstinence at 6‐month follow‐up or more	Study population		RR 1.36 (1.05 to 1.76)	7560 (10 RCTs)	⊕⊕⊕⊝ MODERATE^a,b	‐
	12 per 100	17 per 100 (13 to 22)
*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).  CI: Confidence interval; RCT: Randomized controlled trial; RR: Risk ratio
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

Biomedical feedback in addition to standard smoking cessation treatment in primary care
Patient or population: people who attend primary care and smoke tobacco Setting: primary care (Europe, USA) Intervention: biomedical feedback plus standard smoking cessation support Comparison: standard smoking cessation support
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with placebo	Risk with biomedical feedback
Smoking abstinence at 6‐month follow‐up or more	Study population		RR 1.07 (0.81 to 1.41)	3491 (7 RCTs)	⊕⊕⊝⊝ LOW^a	‐
	10 per 100	11 per 100 (8 to 14)
*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).  CI: Confidence interval; RCT: Randomized controlled trial; RR: Risk ratio
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

Tailored print materials in addition to standard smoking cessation treatment in primary care
Patient or population: people who attend primary care and smoke tobacco Setting: primary care (Europe) Intervention: tailored print materials plus standard smoking cessation support Comparison: standard smoking cessation support
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with placebo	Risk with tailored print materials
Smoking abstinence at 6‐month follow‐up or more	Study population		RR 1.29 (1.04 to 1.59)	15,978 (6 RCTs)	⊕⊕⊕⊝ MODERATE^a	‐
	3 per 100	4 per 100 (4 to 5)
*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).  CI: Confidence interval; RCT: randomized controlled trial; RR: Risk ratio
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

Provider training in addition to standard smoking cessation treatment in primary care
Patient or population: people who attend primary care and smoke tobacco Setting: primary care (Argentina, Canada, Europe, USA) Intervention: provider training plus standard or multicomponent smoking cessation support Comparison: standard or multicomponent smoking cessation support
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with placebo	Risk with provider training
Smoking abstinence at 6‐month follow‐up or more	Study population		RR 1.10 (0.85 to 1.41)	13,685 (7 RCTs)	⊕⊕⊝⊝ LOW^a,b	‐
	5 per 100	6 per 100 (5 to 8)
*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).  CI: Confidence interval; RCT: Randomized controlled trial; RR: Risk ratio
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

Provider incentives in addition to standard smoking cessation treatment in primary care
Patient or population: people who attend primary care and smoke tobacco Setting: primary care (Germany, USA) Intervention: provider incentives plus standard or multicomponent smoking cessation support Comparison: standard or multicomponent smoking cessation support
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with placebo	Risk with provider incentives (provider‐level)
Smoking abstinence at 6‐month follow‐up or more	Study population		RR 1.14 (0.97 to 1.34)	2454 (2 RCTs)	⊕⊝⊝⊝ VERY LOW^a,b	‐
	18 per 100	21 per 100 (17 to 24)
*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).  CI: Confidence interval; RR: Risk ratio; OR: Odds ratio;
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

Search	Query
#28	(#23 AND #24 AND #27) (smoking terms, primary care terms, study terms (no animals))
#27	(#26 NOT #20) (All study terms NOT animals)
#26	(#25 OR #21 OR #22) (Cochrane with eval and clinical)
#25	(#13 OR #14 OR #15 OR #16 OR #17 OR #18 OR #19) (Cochrane Search)
#24	(#8 OR #9 OR #10 OR #11 OR #12) (Primary Care Terms)
#23	(#1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7) (Smoking Terms)
#22	clinical trial
#21	evaluation studies
#20	(animals [mh] NOT humans [mh])
#19	trial [ti]
#18	randomly [tiab]
#17	clinical trials as topic [mesh: noexp]
#16	placebo [tiab]
#15	randomized [tiab]
#14	controlled clinical trial [pt]
#13	randomized controlled trial [pt]
#12	general practitioner*
#11	general practice*
#10	family physician*
#9	primary care
#8	primary health care
#7	tobacco use disorder
#6	tobacco use cessation
#5	smoking/therapy
#4	smoking/prevention and control
#3	smoking cessation
#2	nicotine
#1	tobacco

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Long‐term abstinence (subgrouped by single vs. multicomponent intervention type)	22	18150	Risk Ratio (M‐H, Random, 95% CI)	1.31 [1.10, 1.55]
1.1.1 Adjunctive counseling + standard care vs standard care	17	12852	Risk Ratio (M‐H, Random, 95% CI)	1.43 [1.15, 1.78]
1.1.2 Adjunctive counseling + multicomponent int vs multicomponent int	5	5298	Risk Ratio (M‐H, Random, 95% CI)	1.04 [0.87, 1.23]
1.2 Long‐term abstinence (subgrouped by provider)	22	18150	Risk Ratio (M‐H, Random, 95% CI)	1.31 [1.10, 1.55]
1.2.1 Nurses	11	3214	Risk Ratio (M‐H, Random, 95% CI)	1.20 [0.97, 1.50]
1.2.2 Psychologists & counselors	12	14835	Risk Ratio (M‐H, Random, 95% CI)	1.31 [1.03, 1.66]
1.2.3 Pharmacists	1	101	Risk Ratio (M‐H, Random, 95% CI)	2.38 [0.99, 5.70]
1.3 Long‐term abstinence (subgrouped by mode)	22	18150	Risk Ratio (M‐H, Random, 95% CI)	1.31 [1.10, 1.55]
1.3.1 Face to face	14	11753	Risk Ratio (M‐H, Random, 95% CI)	1.31 [1.06, 1.61]
1.3.2 Telephone	10	6397	Risk Ratio (M‐H, Random, 95% CI)	1.31 [0.98, 1.75]
1.4 Long‐term abstinence (subgrouped by intensity)	22	18150	Risk Ratio (M‐H, Random, 95% CI)	1.31 [1.10, 1.55]
1.4.1 Brief/minimal	6	2533	Risk Ratio (M‐H, Random, 95% CI)	1.42 [1.07, 1.88]
1.4.2 More substantial	18	15617	Risk Ratio (M‐H, Random, 95% CI)	1.28 [1.04, 1.57]
1.5 Advise rates	2	724	Risk Ratio (M‐H, Random, 95% CI)	1.08 [0.93, 1.26]
1.5.1 Adjunctive counseling + standard care vs standard care	1	190	Risk Ratio (M‐H, Random, 95% CI)	1.20 [0.94, 1.54]
1.5.2 Adjunctive counseling + multicomponent int vs multicomponent int	1	534	Risk Ratio (M‐H, Random, 95% CI)	1.02 [0.87, 1.19]
1.6 Assistance rates	5		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
1.6.1 Medication	3	1094	Risk Ratio (M‐H, Random, 95% CI)	1.61 [1.20, 2.15]
1.6.2 Counseling	3	1460	Risk Ratio (M‐H, Random, 95% CI)	1.64 [0.94, 2.88]
1.6.3 Quit date set	1	190	Risk Ratio (M‐H, Random, 95% CI)	1.42 [0.58, 3.44]
1.6.4 Self‐help materials	1	190	Risk Ratio (M‐H, Random, 95% CI)	1.00 [0.70, 1.42]
1.7 Arrange follow‐up support rates	3		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
1.7.1 Adjunctive counseling + standard care vs standard care	3	1718	Risk Ratio (M‐H, Random, 95% CI)	4.65 [1.67, 12.90]
1.8 Quit attempts	3		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
1.8.1 Adjunctive counseling + standard care vs standard care	3	1764	Risk Ratio (M‐H, Random, 95% CI)	1.23 [1.02, 1.49]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Long‐term abstinence (subgrouped by single vs. multicomponent intervention type)	10	7560	Risk Ratio (M‐H, Random, 95% CI)	1.36 [1.05, 1.76]
2.1.1 Cost‐free meds + standard care vs standard care	6	4975	Risk Ratio (M‐H, Random, 95% CI)	1.46 [1.05, 2.03]
2.1.2 Cost‐free meds + multicomponent int vs multicomponent int	4	2585	Risk Ratio (M‐H, Random, 95% CI)	1.21 [0.81, 1.82]
2.2 Quit attempts	3		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
2.2.1 Cost‐free meds + standard care vs standard care	3	2669	Risk Ratio (M‐H, Random, 95% CI)	1.21 [1.02, 1.43]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Long‐term abstinence (subgrouped by type)	7	3491	Risk Ratio (M‐H, Random, 95% CI)	1.07 [0.81, 1.41]
3.1.1 Spirometry	4	2137	Risk Ratio (M‐H, Random, 95% CI)	1.16 [0.72, 1.86]
3.1.2 CO monitoring	1	1040	Risk Ratio (M‐H, Random, 95% CI)	1.15 [0.87, 1.51]
3.1.3 Gene testing for lung cancer risk	1	109	Risk Ratio (M‐H, Random, 95% CI)	0.85 [0.40, 1.80]
3.1.4 CO monitoring & spirometry	1	205	Risk Ratio (M‐H, Random, 95% CI)	0.64 [0.29, 1.40]

*Study characteristics*
Methods	Design: Randomized controlled trial Setting: Primary care clinic, Spain Recruitment: Clinic & community volunteers
Participants	48 people who smoked (excludes 6 dropouts) 65% female, av. age 36, av. cpd 24 ‐ 27 Therapist: unclear, primary care clinic staff
Interventions	Intervention: participants received 4 x 30‐minute sessions of counseling over 4 weeks, which consisted of video, cognitive therapy, social influences and relapse prevention Control: participants received 3 minutes of advice immediately after randomization
Outcomes	Abstinence at 12m (abstinence not defined) Validation: None
Funding Source	Not reported
Author's declarations of interest	Not reported
Notes	Strategy: Adjunctive counseling Level: Patient Comparison type: Single component vs. standard care
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sequence Generation	Unclear risk	No details reported
Allocation concealment	Unclear risk	No details reported
Blinding of outcome assessors All outcomes	High risk	Participants self‐reported smoking status in person or by telephone. Intervention group received greater face‐to‐face contact
Incomplete outcome data All outcomes	Low risk	Loss to follow‐up low and similar between arms: 4/25 control, 2/29 intervention

*Study characteristics*
Methods	Design: Randomized controlled trial Setting: Primary care clinics of Veterans Affairs Medical Centers, USA Recruitment: Invitation letter to primary care patients
Participants	837 people who smoked daily, 26 cpd, av.age 57, 10% F
Interventions	Intervention: participants received behavioral counseling via telephone (7 calls over 2 months) with mailing of smoking cessation medications as clinically indicated. Additional calls were placed over a 12‐month period at the discretion of the counselor Control: participants were mailed self‐help materials and had continued access to clinical smoking cessation services through their Veterans Affairs Medical Center. All of these had referral‐based smoking cessation programs. Program structure varied by site (e.g., number of sessions and group or individual therapy). However, nicotine patches, nicotine gum, and slow‐release bupropion were available at all sites
Outcomes	6m sustained abstinence at 12m Validation: None Measures of provider implementation: Assist‐Meds, Arrange
Funding Source	Department of Veterans Affairs Health Services Research and Development Service grant SUI 99101‐1
Author's declarations of interest	Authors declared that they had no financial conflict of interest.
Notes	Strategy: Adjunctive counseling Level: Patient Comparison type: Single component vs. standard care
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sequence Generation	Low risk	Computer‐generated randomization scheme stratified by primary care facility and blocked within sites
Allocation concealment	Low risk	Computer‐generated
Blinding of outcome assessors All outcomes	High risk	Smoking status was self‐reported. Intervention group had greater face‐to‐face contact.
Incomplete outcome data All outcomes	Low risk	The overall loss to follow‐up was 14.6% (n = 122/838); 13.9% (n = 58/418) in the intervention group and 15.2% (n = 64/420) in the control group were lost to follow‐up at 12 months

*Study characteristics*
Methods	Design: Randomized controlled trial Setting: Inner‐city hospital‐based primary care clinics, UK Recruitment: During routine healthcare visits in primary care clinics
Participants	846 adults who smoked randomized to intervention (n = 406) and control (n = 440), 68.7% female, average age 39, cpd of at least 10
Interventions	Intervention: participants received a 45‐minute counseling session with health educators and follow‐up calls at quit date and 2 weeks, in addition to the standard care (described below) Control: Healthcare professionals received training on smoking cessation guidelines and applying the 5 As Participants received brief advice from their physician and 8 weeks of NRT
Outcomes	7‐day PPA at 12m Validation: Expired CO ≤ 5 ppm
Funding Source	National Institutes of Health, National Institute on Drug Abuse (R01DA010860)
Author's declarations of interest	Authors declared that they had no conflict of interest.
Notes	Strategy: Provider training, adjunctive counseling, cost‐free medication Level: Patient, provider Comparison type: Active vs. active
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sequence Generation	Low risk	The computer used a random number program to assign participants at random to one of two treatment conditions
Allocation concealment	Low risk	Computerised system
Blinding of outcome assessors All outcomes	Low risk	Smoking status was validated by carbon monoxide
Incomplete outcome data All outcomes	High risk	Follow‐up rate was low overall (< 50%) : Intervention: 47.3%, Control: 41.4%

*Study characteristics*
Methods	Design: Randomized controlled trial Setting: 2 health centres in the Basque Health Service, Spain Recruitment: Identified through EMR and sent invitation letter
Participants	320 adults who smoked, 44% female, average age 45
Interventions	Intervention: participants received health advice from a doctor or nurse, as in the other group, plus reinforcement text messages to their mobile phones Control: participants received health advice provided by a doctor or a nurse
Outcomes	Continuous abstinence at 12m Validation: Expired CO < 7 ppm
Funding Source	This study was funded by the Departamento de Industria del Gobierno Vasco of the Basque Country under the 2012 Saiotek funding round (reference number SAIO12‐OA12BF001). This research was also supported by Departamento de Educación, Política Lingüística y Cultura del Gobierno Vasco (IT620‐13)
Author's declarations of interest	Authors declared that they had no conflict of interest
Notes	Strategy: SMS messages Level: Patient Comparison type: Single component vs. standard care
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sequence Generation	Low risk	Computer‐generated sequence
Allocation concealment	Low risk	QUOTE: "the Bioaraba Research Institute assigned patients to one of the two arms of the trial...after receiving the patient randomization form, and hence research nurse did not know about the treatment group until patient allocation"
Blinding of outcome assessors All outcomes	Low risk	Smoking status was validated by carbon monoxide
Incomplete outcome data All outcomes	High risk	The overall loss to follow‐up was 87.5% (n = 280/320); 80.6% (n = 129/160) in the text messaging+health advice group and 94.4% (n = 151/160) in the health advice‐only group were lost to follow‐up at 12 months

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
4.1 Long‐term abstinence (subgrouped by theoretical basis)	6	15978	Risk Ratio (M‐H, Random, 95% CI)	1.29 [1.04, 1.59]
4.1.1 Based on transtheoretical model	2	2470	Risk Ratio (M‐H, Random, 95% CI)	1.52 [1.07, 2.17]
4.1.2 No clear theoretical basis	4	13508	Risk Ratio (M‐H, Random, 95% CI)	1.20 [0.90, 1.61]
4.2 Quit attempts	3	11122	Risk Ratio (M‐H, Random, 95% CI)	1.08 [1.00, 1.17]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
5.1 Long‐term abstinence	7	13685	Risk Ratio (M‐H, Random, 95% CI)	1.10 [0.85, 1.41]
5.1.1 Provider training + standard care vs standard care	5	12011	Risk Ratio (M‐H, Random, 95% CI)	1.11 [0.81, 1.53]
5.1.2 Provider training + multicomponent int vs multicomponent int	2	1674	Risk Ratio (M‐H, Random, 95% CI)	1.04 [0.56, 1.93]
5.2 Asking rates	4	3591	Risk Ratio (M‐H, Random, 95% CI)	1.09 [1.05, 1.13]
5.2.1 Provider training + standard care vs standard care	3	2724	Risk Ratio (M‐H, Random, 95% CI)	1.08 [1.03, 1.12]
5.2.2 Provider training + multicomponent int vs multicomponent int	1	867	Risk Ratio (M‐H, Random, 95% CI)	1.18 [1.06, 1.31]
5.3 Advise rates	4	4112	Risk Ratio (M‐H, Random, 95% CI)	1.12 [1.02, 1.24]
5.3.1 Provider training + standard care vs standard care	2	2438	Risk Ratio (M‐H, Random, 95% CI)	1.21 [0.95, 1.54]
5.3.2 Provider training + multicomponent int vs multicomponent int	2	1674	Risk Ratio (M‐H, Random, 95% CI)	1.05 [0.98, 1.12]
5.4 Assistance rates	5		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
5.4.1 Quit date set	3	3305	Risk Ratio (M‐H, Random, 95% CI)	1.64 [0.86, 3.14]
5.4.2 Self‐help materials	4	4380	Risk Ratio (M‐H, Random, 95% CI)	2.32 [1.16, 4.62]
5.4.3 Medication	2	1674	Risk Ratio (M‐H, Random, 95% CI)	1.15 [0.90, 1.47]
5.4.4 Counseling	1	867	Risk Ratio (M‐H, Random, 95% CI)	1.26 [1.09, 1.45]
5.5 Arrange follow‐up support rates	3	2674	Risk Ratio (M‐H, Random, 95% CI)	1.60 [0.95, 2.69]
5.6 Quit attempts	5	6700	Risk Ratio (M‐H, Random, 95% CI)	1.04 [0.98, 1.10]
5.6.1 Provider training + standard care vs standard care	3	5026	Risk Ratio (M‐H, Random, 95% CI)	1.02 [0.94, 1.10]
5.6.2 Provider training + multicomponent int vs multicomponent int	2	1674	Risk Ratio (M‐H, Random, 95% CI)	1.08 [0.98, 1.19]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
6.1 Long‐term abstinence	2	2454	Risk Ratio (M‐H, Random, 95% CI)	1.14 [0.97, 1.34]
6.1.1 Provider incentives + standard care vs standard care	1	2089	Risk Ratio (M‐H, Random, 95% CI)	1.16 [0.98, 1.37]
6.1.2 Provider incentives + multicomponent int vs multicomponent int	1	365	Risk Ratio (M‐H, Random, 95% CI)	0.85 [0.43, 1.70]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
9.1 Long‐term abstinence	3	2651	Risk Ratio (M‐H, Random, 95% CI)	1.70 [1.27, 2.27]
9.2 Asking rates	2		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
9.3 Assistance rates	2		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
9.3.1 Medication	2		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
9.4 Arrange follow‐up support rates	2	430	Risk Ratio (M‐H, Random, 95% CI)	5.53 [0.41, 73.81]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
10.1 Long‐term abstinence	2	2972	Risk Ratio (M‐H, Random, 95% CI)	1.55 [0.95, 2.52]
10.2 Asking rates	2	2700	Risk Ratio (M‐H, Random, 95% CI)	1.27 [0.98, 1.64]
10.3 Assistance rates	2		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
10.3.1 Medication	2	1321	Risk Ratio (M‐H, Random, 95% CI)	0.83 [0.57, 1.22]
10.3.2 Quit date set	2	1701	Risk Ratio (M‐H, Random, 95% CI)	5.73 [4.19, 7.83]
10.3.3 Self‐help materials	2	2700	Risk Ratio (M‐H, Random, 95% CI)	3.34 [2.54, 4.38]
10.4 Arrange follow‐up support rates	2	1321	Risk Ratio (M‐H, Random, 95% CI)	1.53 [1.15, 2.03]
10.5 Quit attempts	2	2972	Risk Ratio (M‐H, Random, 95% CI)	1.05 [0.96, 1.15]

*Study characteristics*
Methods	Design: 4‐group randomized controlled trial Setting: General practices in West Midlands, UK Recruitment: Mailed invitations to patients of general practices.
Participants	65 practices 2471 people who smoked, 55% F, av.age 41, 20 cpd
Interventions	Intervention 1: participants received self help workbook and three tailored letters Intervention 2: patients received self‐help workbook, three tailored letters, and three telephone calls Intervention 3: patients received self‐help workbook, three tailored letters, and three appointments with a nurse Control: patients received four standard items of self‐help materials
Outcomes	6m sustained abstinence at 12m Validation: Salivary cotinine <14.2 ng/ml
Funding Source	The health authorities of the West Midlands
Author's declarations of interest	Not reported.
Notes	Strategy: Adjunctive counseling + Tailored print materials Level: Patient Comparison type: 1) Single component vs. standard care; 2) multicomponent vs. standard care; 3) active vs. active
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sequence Generation	Unclear risk	No details reported
Allocation concealment	Low risk	QUOTE: "Questionnaires were read optically and the data transferred automatically to the Access database that performed the minimization and controlled the contacts. There was no reason and no way that the clerical assistant running the database could alter the questionnaire reading schedule, which would have altered the allocation of particular individuals"
Blinding of outcome assessors All outcomes	Low risk	Smoking abstinence was biochemically validated
Incomplete outcome data All outcomes	Low risk	The overall loss to follow‐up was 39.7% (n = 981/2471); 40.0% (n = 273/683) in the manual group, 43.6% (n = 299/685) in the phone group, 49.2% (n = 203/413) in the nurse group, and 29.9% (n = 206/690) in the control group were lost to follow‐up at 12 months

*Study characteristics*
Methods	Design: Cluster‐randomized controlled trial with 2 active comparators Setting: Primary care practices, Australia Recruitment: Conducted in clinic, by GP or practice staff
Participants	1039 adults who smoked, 55.4% female, 17 cpd 45 providers
Interventions	Common components in both groups: GPs were instructed to adhere to the National Guidelines, including a brief assessment of readiness to quit and if relevant, to deal with use of pharmacotherapy Intervention 1: in‐practice management GPs were encouraged to provide people who smoked with additional information and help to stop smoking. GPs were not precluded from recommending external assistance or from referring participants to the quitline if this was their clinical practice Intervention 2: referral GPs were encouraged to offer people who smoked with any interest in quitting referral to the quitline and were provided with a brochure on quitline services. Participants received an introductory call/letter from quitline, followed by up to 2 pre‐quitting and 4 post‐quitting telephone counseling sessions
Outcomes	≥10m sustained abstinence at 12m Validation: None Quit attempts Measures of provider implementation: Assist, Arrange
Funding Source	National Health and Medical Research Council (284346)
Author's declarations of interest	QUOTE: "RB, JB and NB are employees of The Cancer Council Victoria that runs the quitline service used in this study. None are involved in day to day operations of the service"
Notes	Strategy: Adjunctive counseling Level: Patient Comparison type: Single component vs. standard care
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sequence Generation	Low risk	QUOTE: "GPs were randomised by computer prior to their attendance at an education session"
Allocation concealment	Low risk	Computerised system
Blinding of outcome assessors All outcomes	High risk	Smoking status was self‐reported and not validated biochemically. The amount of contact differed between trial arms
Incomplete outcome data All outcomes	Low risk	Attrition rates were under 50% and similar across the groups. The overall loss to follow‐up was 33.6% (n = 349/1039); 32.0% (n = 233/728) in the referral group and 37.3% (n = 116/311) in the in‐practice group were lost to follow up at 12 months
Recruitment bias (cluster RCTs only)	Low risk	Participants were affiliated with the practice before randomization. QUOTE: "Patients who presented for any reason who were current smokers...were eligible for recruitment"; "Method of patient recruitment did not differ by condition (P=0.79)"
Balanced baseline characteristics? (cluster RCTs only)	Low risk	No significant differences between groups
Adjustment for clustering in analysis? (cluster RCTs only)	Low risk	Statistical analyses were performed with Stata, controlling for practice as a clustering variable QUOTE: "In order to take into account the correlated nature of the data and repeated measures over time, generalized estimating equations were used for a final analysis of outcomes"

*Study characteristics*
Methods	Design: Cluster‐randomized trial Setting: 22 primary care clinics in South Carolina, USA Recruitment: patients identified at routine visits
Participants	1245 adults who smoked, 61% female, average age 50.7, average cpd 15
Interventions	Intervention: cessation advice and brochure with information on quitline, plus a 2‐week supply of both nicotine patch and lozenge, with minimal instructions on use Control: cessation advice and brochure with information on quitline Training given to providers was based on study procedures and standard care
Outcomes	7‐day PPA at 6m Validation: none Quit attempts
Funding Source	National Institute on Drug Abuse (R01 DA 021619), with additional research support through NIH UL1 TR001450 and K23 DA 045766
Author's declarations of interest	Some authors have received consulting honoraria from Pfizer (does not produce NRT)
Notes	Strategy: Cost‐free medication Level: Patient Comparison type: Single component vs. standard care
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sequence Generation	Low risk	Stratified randomization lists created at the outset of the study
Allocation concealment	Low risk	Accessible only to the study statistician
Blinding of outcome assessors All outcomes	Low risk	Smoking status was not validated, but both groups had minimal contact, with no difference in the face‐to‐face contact between the arms
Incomplete outcome data All outcomes	Low risk	At patient level, the number of participants lost was 41.1% (512/1245); 40.3% (263/652) in the standard care group and 42.0% (249/593) in the intervention group, no significant differences between groups, intention‐to‐ treat used to analyze data
Recruitment bias (cluster RCTs only)	Low risk	22/24 clinics approached agreed to participants. Patient participants were affiliated with the practice before randomization. 6 eligible patients did not enrol
Balanced baseline characteristics? (cluster RCTs only)	High risk	Quote: "several baseline variables differed significantly between groups…" and adjusted for in the analysis
Adjustment for clustering in analysis? (cluster RCTs only)	Low risk	adjustment for clustering was conducted

*Study characteristics*
Methods	Design: Randomized controlled trial with 3 active trial arms Setting: Rural primary care practices, Kansas, USA
Participants	726 adults who smoked >10 cpd, randomized to intervention (n = 482) and control (n = 244), 41.5% M; av.age 47.2, 24 cpd
Interventions	Common component in all groups: offer of free pharmacotherapy Intervention 1: pharmacotherapy only At baseline and 6, 12,and 18 months, participants received a mailed offer of free pharmacotherapy that consisted of either 6‐weeks of nicotine patch (21 mg/d) or 7‐weeks of sustained‐release bupropion (150 mg twice daily) Intervention 2: moderate‐intensity disease management Participants received an offer of free pharmacotherapy (as above) with educational support and 2 telephone‐based counseling sessions every 6 months Intervention 3: high‐intensity disease management Participants received an offer of free pharmacotherapy (as above) with educational support and 6 telephone‐based counseling sessions every 6 months
Outcomes	7‐day PPA at 24m Validation: Salivary cotinine level < 15 ng/mL in a mailed saliva sample. Because of resistance by participants to providing salivary samples at month 12, validation by proxy report from a significant other at month 24 was used for quitters who did not return a salivary sample. The validated quit rate at 24m is a mixture of the 2 approaches
Funding Source	National Cancer Institute (grant R01‐101963). Study medication was provided by GlaxoSmithKline
Author's declarations of interest	Authors declared that they had no conflict of interest
Notes	Strategy: Adjunctive counseling + cost‐free medication Level: Patient Comparison type: Active vs. active (isolating adjunctive counseling)
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sequence Generation	Low risk	QUOTE: "randomization occurred at the participant level. A computer‐generated random‐number table was used to generate allocation cards in blocks of 24, with allocation equally distributed across treatment groups"
Allocation concealment	Low risk	QUOTE: "to conceal allocation, we placed these cards in sequentially numbered, opaque, sealed envelopes"
Blinding of outcome assessors All outcomes	Unclear risk	Smoking status was biochemically validated, but there was a low return rate and so proxy report was also used
Incomplete outcome data All outcomes	Low risk	The overall loss to follow‐up was 17.3% (n = 130/750); 13.2% (n = 33/250) in the PM group, 20.0% (n = 50/249) in the MDM group and 18.7% (n = 47/251) in the HDM group were lost to follow‐up at 24 months

*Study characteristics*
Methods	Design: 3‐group cluster‐randomized controlled trial Setting: Primary care, The Netherlands Recruitment: Patients with COPD identified from medical records in participating practices
Participants	667 patients with COPD (148 control, 243 intervention 1, 276 intervention 2), age > 35 years, av.age 60, 16 cpd, 49% M
Interventions	Intervention 1: counseling strategy, recommendation of NRT • The general practice team received a 4‐hour group training session about chronic obstructive pulmonary disease and smoking cessation, and 3 visits by an outreach visitor for additional individual support • Participants received a leaflet especially developed for people with chronic obstructive pulmonary disease who smoked, a videotape, self‐efficacy enhancing information, information about NRT, proactive telephone calls Intervention 2: counseling strategy, recommendation of NRT, advice to use bupropion‐SR • The general practice team received a 4‐hour group training session about chronic obstructive pulmonary disease and smoking cessation, and 3 visits by an outreach visitor for additional individual support • Participants received a leaflet especially developed for people with chronic obstructive pulmonary disease who smoked, a videotape, self‐efficacy enhancing information, information about NRT, proactive telephone call, and advice to use bupropion‐SR Control: usual care consisting of periodic regular check‐ups and chronic obstructive pulmonary disease information
Outcomes	PPA at 12m Validation: Urinary cotinine < 50 ng/mL
Funding Source	Financed by the Dutch Asthma Foundation, Netherlands Organization for Health Research and Development (ZonMW), and Pharmacia
Author's declarations of interest	Authors declared that they had no conflict of interest
Notes	Strategy: Provider training, outreach facilitation, adjunctive counseling Level: Patient + Provider + Practice Comparison type: Multicomponent vs. SC
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sequence Generation	Unclear risk	No details reported
Allocation concealment	Unclear risk	No details reported
Blinding of outcome assessors All outcomes	Low risk	Smoking status was biochemically validated
Incomplete outcome data All outcomes	Low risk	Attrition rates were under 50% and similar between groups. The overall loss to follow‐up was 18.7% (n = 130/697); 3.6% (n = 9/252) in 'counseling + NRT' group, 5.2% (n = 15/291) in 'counseling+NRT+prescription of bupropion' group, 3.9% (n = 6/154) in the control group were lost to follow‐up at 1 year.
Recruitment bias (cluster RCTs only)	Low risk	Participants were affiliated with the practice before randomization.
Balanced baseline characteristics? (cluster RCTs only)	Low risk	Groups were balanced
Adjustment for clustering in analysis? (cluster RCTs only)	Low risk	QUOTE: "We used multilevel analyses to test treatment effects because of the study's hierarchical structure..."

PERMALINK

Strategies to improve smoking cessation rates in primary care

Nicola Lindson

Gillian Pritchard

Bosun Hong

Thomas R Fanshawe

Andrew Pipe

Sophia Papadakis

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Summary of findings

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Primary outcomes

Secondary outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Methods for categorizing details of intervention

Assessment of risk of bias in included studies

Measures of treatment effect

Unit of analysis issues

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Summary of findings and assessment of the certainty of the evidence

Results

Description of studies

Results of the search

1.

Included studies

Participants

Interventions and comparators

Outcomes

Excluded studies

Risk of bias in included studies

2.

Allocation

Blinding (detection bias)

Incomplete outcome data

Recruitment bias due to recruitment of participants to clusters after allocation (cluster‐RCTs only)

Unbalanced baseline characteristics (cluster‐RCTs only)

Statistical adjustment to account for potential correlation effects within clusters (cluster‐RCTs only)

Other potential sources of bias

Effects of interventions

Summary of findings 1. Adjunctive counseling in addition to standard smoking cessation care in primary care.

Summary of findings 2. Cost‐free medications used in addition to standard care in primary care.

Summary of findings 3. Biomedical feedback in addition to standard smoking cessation treatment in primary care.

Summary of findings 4. Tailored print materials in addition to standard smoking cessation treatment in primary care.

Summary of findings 5. Provider training in addition to standard smoking cessation treatment in primary care.

Summary of findings 6. Provider incentives in addition to standard smoking cessation treatment in primary care.

Adjunctive counseling (patient‐level strategy)

1.1. Analysis.

1.2. Analysis.

1.3. Analysis.

1.4. Analysis.

1.5. Analysis.

*Study characteristics*
Methods	Design: 3‐group randomized controlled trial Setting: 6 general practices in Aberdeen, UK Recruitment: Mailed lettered to patients identified in EMR
Participants	2553 people who smoked aged 17 ‐ 65, av. age not reported, av. cpd not reported
Interventions	Intervention 1: participants received an untailored letter on smoking cessation Intervention 2: participants received a tailored letter on smoking cessation Control: participants received a letter thanking them for participation and informing them that they would receive material at the end of the study (either a tailored or a non‐tailored letter)
Outcomes	7‐day PPA at 6m Validation: Salivary cotinine, cut‐off not reported (only completed in 3.5% of participants)
Funding Source	The Chief Scientist Office, Scottish Executive Health Department, with additional funding from the Engineering and Physical Sciences Research Council. The Health Economics Research Unit is funded by the Chief Scientist Office
Author's declarations of interest	Authors declared that they had no conflict of interest
Notes	Strategy: Tailored print materials Level: Patient Comparison type: Single component vs. standard care
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sequence Generation	Low risk	Computer‐generated random numbers were used
Allocation concealment	Unclear risk	No details reported
Blinding of outcome assessors All outcomes	Low risk	Salivary cotinine, cut‐off not reported (only completed in 3.5% of participants). Face‐to‐face contact was minimal in all groups
Incomplete outcome data All outcomes	Low risk	The overall loss to follow‐up was 23.6% (n = 615/2610); 24.5% (n = 213/870) in the tailored letter group, 27.2% (n = 236/869) in the standard letter group and 19.1% (n = 166/871) in the control group were lost to follow‐up at 6 months

*Study characteristics*
Methods	Design: Clustered randomized controlled trial with 3 active comparators Setting: Primary care in North‐Eastern Germany Recruitment: Practices contacted by research team and invited to participate
Participants	151 practices 3086 participants, 43% F, av.age 40, av. cpd not reported
Interventions	Intervention 1: brief advice with desktop resource Participants received a 10‐minute brief advice which incorporated elements of health behavior change counseling and were provided with self‐help materials. Physicians received a summary sheet of smoking‐related characteristics to prompt counseling and also a desktop resource with a flow chart illustrating the elements of counseling and general communication strategies Intervention 2: tailored letters Participants received 2 tailored letters, based on their answers to 2 questionnaires. Letters were given while participant was in the practice Intervention 3: combination Participants received both brief advice from the physician and a tailored letter. The same self‐help manuals used in the other conditions were provided to the participants
Outcomes	6m prolonged abstinence at 12m Validation: None Provided some data on intervention 'reach' but it was not possible to classify this into our provider implementation outcomes
Funding Source	The German Federal Ministry of Research and EducationThe Social Ministry of the State of Mecklenburg‐Vorpommern The German Research Foundation
Author's declarations of interest	Authors declared that they had no conflict of interest
Notes	Strategy: Tailored materials, flow sheet Level: Patient, practice Comparison type: Active vs. active (2 comparisons: int 1 vs. int 3 & int 2 vs. intervention 3)
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sequence Generation	Unclear risk	No details reported
Allocation concealment	Unclear risk	No details reported
Blinding of outcome assessors All outcomes	High risk	Smoking status self‐report, and contact differed between trial arms
Incomplete outcome data All outcomes	Low risk	Attrition rates were under 50% and the difference among groups was less than 20%. At practice level, no practice was reported to have been lost to follow up (n = 0/151), At participant level, the overall loss to follow‐up was 26.8% (n = 863/3215); 30.6% (n = 189/618) in the brief advice group, 22.3% (n = 331/1484) in the tailored letters group, and 30.8% (n = 343/1113) in the combination group were lost to follow‐up at 12 months
Recruitment bias (cluster RCTs only)	Unclear risk	Participants were affiliated with the practice before randomization
Balanced baseline characteristics? (cluster RCTs only)	Low risk	QUOTE: "There were no significant differences in the characteristics of the participating practices and practitioners between study groups"
Adjustment for clustering in analysis? (cluster RCTs only)	Low risk	QUOTE: "For all analyses at the patient level, we took clustering within practices into account using sample survey methods in STATA 10.1"

*Study characteristics*
Methods	Design: Cluster‐ randomized controlled trial Setting: Healthcare centres in Area 11 of the Spanish Madrid Health System, Spain Recruitment: Participants selected from computerized clinic records
Participants	405 nurses and 425 doctors from the 35 clinics 5910 adults who smoked. av.age 43, 53% F, av. cpd not reported
Interventions	Intervention: professionals received 4 x 90‐minute training sessions on smoking cessation Control: professionals were offered the training after finishing the follow‐up period
Outcomes	Continuous abstinence at 6m Validation: Salivary cotinine < 13 ng/ml
Funding Source	Spanish Public Health Investigations Fund, with no role in the study
Author's declarations of interest	QUOTE: "Two authors (FJA and EOE) work as directors of Cantabria University Tobacco Control Master"
Notes	Strategy: Provider training Level: Provider Comparison type: Single component vs. standard care
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sequence Generation	Low risk	QUOTE: "...randomly selected 15 patients for each quota by using number combinations generated with EPIDAT 2.0 software"
Allocation concealment	Low risk	QUOTE: "An independent research assistant assigned the 35 health care centers randomly using SPSS v.12 software...". No further detail provided.
Blinding of outcome assessors All outcomes	Low risk	Abstinence biochemically validated
Incomplete outcome data All outcomes	Unclear risk	Not reported
Recruitment bias (cluster RCTs only)	Low risk	Participants were affiliated with the practice before randomization
Balanced baseline characteristics? (cluster RCTs only)	Low risk	QUOTE: "No significant differences were found between both groups with respect to profession, age, sex, professional experience, percentage of smokers, and previous training in treating tobacco addiction"
Adjustment for clustering in analysis? (cluster RCTs only)	High risk	Not apparent

*Study characteristics*
Methods	Design: Fractional factorial screening experiment Setting: Primary care clinics in southern Wisconsin, USA Recruitment: participants were recruited from 11 primary care clinics. During clinic visits, clinical care staff were prompted by electronic health record technology to invite people identified as smoking to participate in the study
Participants	637 participants, 55% F, av. age 45.8, average cpd not reported
Interventions	Intervention 1. Pre‐quit nicotine patch Half the participants were assigned to the active condition and received 14 mg patches for the 3 weeks prior to the TQD, while the other half did not receive prequit patches Intervention 2. Prequit nicotine gum Participants in the active condition received 2 mg nicotine gum for the 3 weeks prior to the TQD (≥ 9 pieces of gum/day, 1 piece/1 – 2 hours); the other half did not. Participants who received both Prequit Patch and Gum were told to use at least 5 pieces/day of gum, unless such use produced adverse effects Intervention 3. Preparation counseling Participants in the active condition received 3 x 20‐min counseling sessions prior to the TQD, focused on coping skills, reduction, and making practice quit attempts,while the other half of participants did not. The sessions 3 weeks and 1 week before the TQD were in‐person, and the week‐2 session was over the phone Intervention 4. In‐person counseling Participants in the intensive condition received 3 x 20‐min face‐to‐face counseling sessions: 1 week pre‐TQD, on the TQD, and at week 1. Sessions focused on skill building and intra‐treatment social support. Participants assigned to the minimal level received 1 x 3‐min in‐person session at Week‐1 Intervention 5. Phone counseling Participants in the intensive condition received 3 x 15‐min phone sessions (TQD, Days 2 and 10), focused on coping skills, avoiding smoking cues, and intra‐treatment social support. Participants assigned to the minimal condition received 1 x 10‐min session on the TQD. Thus, all participants received someTQD phone counseling Intervention 6. Extended medication All participants received combination NRT (nicotine patch + nicotine gum) starting on their TQD. Half were assigned to receive 8 weeks of patches and 8 weeks of nicotine gum. The other half received 16 weeks of patches and 16 weeks of gum
Outcomes	7‐day PPA at 6m Validation: None Quit attempts
Funding Source	Grants from the National Cancer Institute to the University of Wisconsin Center for Tobacco Research and Intervention and by the Wisconsin Partnership Program. Dr. Collins is also supported by NIH grants
Author's declarations of interest	Authors declared that they had no conflict of interest
Notes	Strategy: Adjunctive counseling (preparation phase and cessation phase, in person and via telephone), cost‐free medications (pre‐ and post‐quit, gum and patch) Level: Patient Comparison types: Active vs. active Due to the complexity of the intervention components and combinations of intervention components tested, it was impossible to include this study in any of our meta‐analyses. Instead we describe the authors' conclusions narratively, in supplementary table 3.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sequence Generation	Low risk	Database that used stratified permuted block randomization
Allocation concealment	Unclear risk	No details reported
Blinding of outcome assessors All outcomes	High risk	Smoking status self‐report. Different contact between trial arms
Incomplete outcome data All outcomes	Low risk	The overall loss to follow‐up was 41.3% (n = 263/637); the number lost to follow‐up ranged from 36% to 46% across the 6 factors at 6 months

*Study characteristics*
Methods	Design: 3‐group cluster randomized controlled trial Setting: Primary care, Denmark Recruitment: Practices recruited by mailed invitation letter. All patients seen by GP were registered
Participants	1518 adults who smoked, 62.6% F, av.age 48, 17 cpd
Interventions	Intervention 1: GPs were instructed to briefly and freely talk about smoking with all people who smoked and refer those people who were motivated to quit to a group‐based smoking cessation counseling. Intervention 2: GPs were instructed to briefly and freely talk about smoking with all people who smoked and refer all those motivated to quit to an Internet‐based smoking cessation program Control: GPs were instructed to give smoking cessation advice and assistance to quit as they used to (not necessarily to all people who were smoking). The control group did not have any special program, beyond what is known from a national survey on Danish GPs ultimo 2004. In this study, the control group only registered whether they discussed smoking with the participant or not and the time consumed by counseling
Outcomes	7‐day PPA at 12m Validation: Urinary cotinine < 200 ng/ml
Funding Source	Danish Centre for Evaluation and Health and Technology Assessment
Author's declarations of interest	Not reported
Notes	Strategy: Adjunctive counseling + Internet program Level: Patient Component type: Single component vs. standard care (different single components tested in different intervention arms)
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sequence Generation	Low risk	QUOTE: "The GPs were pre‐randomised at the Research Centre by a computer generated list to one of the three groups"
Allocation concealment	Low risk	QUOTE: "The GPs were pre‐randomised at the Research Centre by a computer generated list to one of the three groups"
Blinding of outcome assessors All outcomes	Low risk	Smoking status biochemically validated
Incomplete outcome data All outcomes	High risk	Overall, 50.2% (n = 758/1518) of participants were lost to follow‐up at 12 months. No further details by group were reported
Recruitment bias (cluster RCTs only)	Low risk	Participants were affiliated with the practice before randomization
Balanced baseline characteristics? (cluster RCTs only)	Low risk	QUOTE: "There were no significant differences between the groups in terms of the number of patients seen or smokers included"
Adjustment for clustering in analysis? (cluster RCTs only)	Low risk	No additional effect on self‐reported 1‐year abstinence rates of either referral to group‐ based SC counseling was found in cluster analyses

*Study characteristics*
Methods	Design: Randomized controlled trial Setting: Primary care clinic, Spain Recruitment: GPs recruited patients in practice
Participants	125 people who smoked daily, attending clinic, motivated to make a quit attempt but not interested in using pharmacotherapy, av.age ~ 40, av.cpd 19, 52% F (intervention), 62% F (control)
Interventions	Intervention: participants received brief advice from their doctor at recruitment, an appointment with clinic nurse 7 days before TQD, on TQD, 1w, 1m, 2m, 3m. Control: participants received brief advice only
Outcomes	12m sustained abstinence at 24m Validation: Expired CO < 8 ppm
Funding Source	Not reported
Author's declarations of interest	Not reported
Notes	Strategy: Adjunctive counseling Level: Patient Comparison type: Single component
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sequence Generation	Unclear risk	QUOTE: "the patients recruited were randomised, according to the clinic from which they came, to the group that received...". No further detail.
Allocation concealment	Unclear risk	No details reported
Blinding of outcome assessors All outcomes	Low risk	Smoking abstinence was biochemically validated
Incomplete outcome data All outcomes	High risk	The overall loss to follow‐up was 42.4% (n = 53/125); 75% (n = 45/60) in the intervention group and 12.3% (n = 8/65) in the control group were lost to follow‐up at 2 years

*Study characteristics*
Methods	Design: 4‐group cluster‐randomized controlled trial Setting: Primary care in the Rhine–Neckar region, Germany Recruitment: Providers recruited patients in practice
Participants	587 adults aged 36 – 75 years who smoked at least 10 cigarettes/day, av. age not reported, av. cpd not reported
Interventions	Intervention 1: • General medical practitioners received a 2‐hour cost‐free group tutorial in methods of promoting smoking cessation. The general medical practitioners were assured a financial remuneration of EUR 130 after study completion for each study participant they recruited who was abstinent at 12 months follow‐up Intervention 2: • General medical practitioners received a 2‐hour cost‐free group tutorial in methods of promoting smoking cessation • Participants were reimbursed up to EUR 130 for the purchase of nicotine replacement therapy or bupropion for up to 12 months. Intervention 3: • General medical practitioners received a 2‐hour cost‐free group tutorial in methods of promoting smoking cessation. General practitioners were assured a financial remuneration of EUR 130 after study completion for each study participant they recruited who was abstinent at 12 months follow‐up • Participants were reimbursed up to EUR 130 for the purchase of nicotine replacement therapy or bupropion for up to 12 months Control: usual care. No further details reported
Outcomes	6m sustained at 12m Validation: Salivary cotinine <15 ng/ml
Funding Source	Funded by the German Ministry of Education andResearch (Bundesministerium für Bildung und Forschung), project number01EB0113, within the context of the Baden–Wurttemberg Research Network on Addiction (project 01EB0113)
Author's declarations of interest	Authors declared that they had no competing interests.
Notes	Strategy: Provider training + Provider incentive + Cost‐free medication Level: Provider + Patient Comparison type: Active vs. active (isolating cost‐free medications & provider incentive) & multicomponent vs. standard care
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Sequence Generation	Low risk	Computer‐generated random sequence used
Allocation concealment	Low risk	QUOTE: "Randomisation was performed centrally at the German Center for Research on Ageing, Heidelberg, Germany...". Computer‐generated random sequence used
Blinding of outcome assessors All outcomes	Low risk	Smoking status validated biochemically
Incomplete outcome data All outcomes	Low risk	At participant level, the overall loss to follow‐up was 15.0% (n = 88/588); 19.7% (n = 15/76) in the usual care group, 15.1% (n = 22/146) in the TI group, 15.9% (n = 23/145) in the TM group, 12.7% (n = 28/221) in the TM+TI group were lost at 12‐month follow‐up
Recruitment bias (cluster RCTs only)	Low risk	Participants were affiliated with the practice before randomization
Balanced baseline characteristics? (cluster RCTs only)	High risk	In arms TM and TI+TM, the proportion of participants in the pre‐contemplation stage was lower, and the proportion of participants in both the contemplation and preparation stages were higher than in the usual care and TI arms (P < 0.001)
Adjustment for clustering in analysis? (cluster RCTs only)	Low risk	QUOTE: "The effect of both interventions on smoking abstinence at 12 months follow‐up was assessed simultaneously in a mixed logistic regression model accounting for cluster randomisation ‐ that is, including a random effect for medical practice in the model..."

Study	Reason for exclusion
Adair 2013	Addressed multiple risk factors. No specific intervention strategy for smoking cessation
Adam 2019	Outcomes: Cessation not reported
Agarwal 2018	Cessation outcome not reported. (Main outcome: quit attempts)
An 2008	Follow‐up < 6 months
Andrews 2001	Cessation outcome not reported.
Aveyard 2007	Compares 2 active interventions ‐ unable to isolate a single component
Bachmann 2019	Outcomes: Cessation not reported
Bakkevig 2000	Intervention not conducted in primary care
Bentz 2007	Follow‐up < 6 months
Bosworth 2008	Outcomes of interest not reported (Main outcome: Medication adherence and improvement of hypertension‐related health behaviors)
Burke 1993	Only evaluated the efficacy of pharmacotherapy
Butler 1999	Counseling performed by general practitioners
Carey 2016	Follow‐up < 6 months
Cheung 2019	Intervention not conducted in primary care
Cockburn 1992	Intervention relates to marketing strategies for smoking cessation programs
Cohen 2011	Outcomes of interest not reported (Main outcome: Change in proportion of participants achieving target glycemic and cardiac risk factor goals)
Coma 2019	Outcomes: Cessation not reported
Conger 1987	Only intervention was advice from GPs
de Ruijter 2018	Did not assess smoking cessation
Dey 1999	Follow‐up < 6 months
Dickinson 2013	Addressed multiple risk factors. Tobacco users could choose to not participate in the smoking cessation part of the intervention
Dignan 2019	Intervention not conducted in primary care
Drexel 2011	Outcomes of interest not reported (Main outcome: Provision of evidence based COPD care)
Dubey 2006	Follow‐up < 6 months
Efraimsson 2008	Follow‐up l< 6 months
Eikelenboom 2013	Addressed multiple risk factors. No specific intervention strategy for smoking cessation
Emery 2019	No specific intervention strategy for smoking cessation
Emmons 2014	Outcomes of interest not reported (Main outcome: Multiple risk behavior score)
Engle 2019	Intervention tests only simple counseling and medication
Escortell‐Mayor 2020	Multiple risk factor study
Etter 2000	Follow‐up < 6 months
Felton 2019	Outcomes: Cessation not reported
Ferketich 2014	Follow‐up < 6 months
Ferrer 2009	Multiple risk factor study
Fiore 2019	Does not report on a cessation outcome.
Flocke 2014	Follow‐up < 6 months
Folz 2016	Smoking was a secondary outcome and very few people who smoked were involved ‐ 9 in intervention group and 1 in control group
Frank 2004	Follow‐up < 6 months
Fu 2015	Outcomes of interest not reported (Main outcome: Perceived skill in use of 5As and confidence in addressing smoking cessation)
Fulton 2019	Intervention not conducted in primary care
Gerbert 2003	Outcomes of interest not reported (Main outcome: Acceptability of video‐doctor program)
Gilbert 1989	GP delivered counseling. NRT was not cost‐free.
Gilbert 1992	GP delivered follow‐up counseling compared to no follow‐up does not meet our criteria for adjunctive counseling
Gilbert 2007	Follow‐up < 6 months
Gilbody 2019	Intervention tests simple counseling and medication.
Godycki‐Cwirko 2014	Addressed multiple risk factors. No specific intervention strategy for smoking cessation
Green 2020	Does not report on a cessation outcome
Grischott 2019	Tests simple counseling intervention
Groner 2000	The participants of the study were not patients of the general practitioner, but rather the people accompanying the patient
Hall 2003	Outcomes of interest not reported (Main outcome: Readiness to quit)
Harding 2019	Does not report on a cessation outcome
Haug 1994	Only brief advice intervention
Houston 2015	Compares 2 active interventions ‐ unable to isolate a single component
Hughes 1981	Not conducted in primary care
Humphris 2004	Outcomes of interest not reported (Main outcome: Knowledge of oral cancer)
Imperial Cancer Research Fund GP Research Group	Only evaluated the efficacy of pharmacotherapy
Javitz 2004	Compares 2 active interventions ‐ unable to isolate a single component
Jennings 2014	Follow‐up < 6 months
Jolly 2017	Not a smoking cessation intervention
Kalkhoran 2016	Cessation outcome not reported
Kalkhoran 2019	Counseling intervention tested, but no cluster randomization
Kamstrup‐Larsen 2019	Tests simple counseling intervention only.
Karner 2012	No smoking cessation intervention tested
Kastaun 2021	Compared 2 types of health provider training head to head. Did not allow for separation of effect of provider training
Kennedy 2019	Not a smoking cessation intervention
Kim 2020	Does not report on a cessation outcome
Kirkman 1994	Multiple risk factor study
Knight 1989	Outcomes of interest not reported (Main outcome: Predictors of quitting smoking)
Krones 2010	Outcomes of interest not reported (Main outcome: Validity of the Theory of Planned Behavior in a decision aid)
Kruse 2020	Follow‐up < 6 months
Lasser 2013	Outcomes of interest not reported (Main outcome: Engagement in smoking cessation treatment)
Leung 2019	Tests simple pharmacotherapy only.
Liang 2019	Tests simple counseling intervention only
Liebmann 2019	Adjunctive counseling intervention tests but not cluster‐randomized
Linder 2009	Follow‐up < 6 months
Lycett 2010	All participants had identical treatment to stop smoking
Machline‐Carrion 2019	Does not include a cessation outcome.
Mahapatra 2019	Not an RCT
Markun 2018	Smoking cessation was never an intended outcome
McAlister 2009	Multiple risk factor study
McEwen 2002	Follow‐up < 6 months
McGrath 2014	Follow‐up < 6 months
McPhee 1991	Cessation outcome not reported.
McRee 2005	Cessation outcome not reported. Main outcome:
McRobbie 2008	Follow‐up < 6 months
Mehring 2014	Follow‐up < 6 months
Minian 2019	Not a smoking cessation intervention
Muckelbauer 2015	Multiple risk factor study
Naughton 2014	Compares 2 active interventions ‐ unable to isolate a single component
NCT01072422	follow‐up < 6 months
NCT03221010	follow‐up < 6 months
NCT04200534	Cessation measured < 6 months
NCT04316260	Cessation measured < 6 months
Neuner‐Jehle 2013	Cessation outcome not reported.
Nilsson 1996	Follow‐up < 6 months
Ojedokun 2013	Follow‐up < 6 months
Papadakis 2013	Follow‐up < 6 months
Parchman 2019	Multiple risk factor study
Peckham 2019	Not conducted in primary care
Peprah 2019	Multiple risk factor study
Persai 2020	Does not report on a cessation outcome
Persell 2013	Outcomes of interest not reported (Main outcome: LDL Cholesterol levels)
Pieterse 2001	GP delivered counseling
Piper 2003	Pre‐post evaluation, not an RCT study design
Prabhakaran 2019	No smoking cessation intervention tested
Prochaska 2005	Multiple risk factor study
Prokhorov 2010	Follow‐up < 6 months
Redfern 2014a	Addressed multiple risk factors. No specific intervention strategy for smoking cessation
Richmond 1985	GP delivered counseling.
Richmond 1998	Cessation outcome not reported.
Richter 2015	Compares 2 active interventions ‐ unable to isolate a single component
Rigotti 2011	Follow‐up < 6 months
Robson 1989	Follow‐up < 6 months
Rodriguez Alverez 2008	Compares 2 active interventions ‐ unable to isolate a single component
Roski 1998	Cessation outcome not reported.
Rosser 1992	Follow‐up < 6 months
Rothemich 2008	Follow‐up < 6 months
Rothemich 2010	Follow‐up < 6 months
Sanders 1989	Only brief advice intervention
Satterfield 2018	Follow‐up < 6 months, cessation outcome not reported
Schwartz 2015	Multiple risk factor study
Sejourne 2010	Outcomes of interest not reported (Main outcome: Readiness to quit)
Senesael 2013	Addressed multiple risk factors. No specific intervention strategy for smoking cessation
Shaughnessy 1987	Couseling only on use of medications
Sheffer 2012	Cessation outcome not reported.
Shelley 2016	Did not set out to measure smoking abstinence
Sherman 2017	Compares 2 active interventions ‐ unable to isolate a single component
Silveira 2019	Simple counseling intervention tested only
Silverman 2004	Cessation outcome not reported.
Slama 1990	GP delivered counseling
Smit 2010	Deviated from protocol, not set in primary care
Sperl‐Hillen 2018	Multiple risk factor study
Stratelis 2006	Compares 2 active interventions ‐ unable to isolate a single component
Strecher 1994	Follow‐up < 6 months
Taylor 2019	Did not occur in primary care
Thompson 1988	GP delivered counseling
Thompson 2015	Follow‐up < 6 months
Valdivieso‐Lopez 2013	Study population is pregnant women
Velasquez 2014	Study population is pregnant women
Vidrine 2013	Follow‐up < 6 months
Vogt 2009	Does not report on abstinence outcome.
Voogdt‐Pruis 2011	Outcomes of interest not reported (Main outcome: Number of lifestyle and medical interventions)
Vorderstrasse 2013	Not a smoking cessation intervention
Waage 1997	No smoking cessation strategy conducted in primary care
Wadland 2007	Does not report on abstinence outcome.
Weingarten 1989	Follow‐up < 6 months
West 1998	Follow‐up < 6 months
Wilson 1982	GP delivered follow‐up counseling
Woollard 1995	Outcomes of interest not reported (Main outcome: Blood pressure)
Yalcin 2014	Only evaluating different intensities of counseling
Yingst 2018	Follow‐up < 6 months
Young 2002a	Follow‐up < 6 months
Young 2002b	Follow‐up < 6 months
Ziyash 2019	Tests simple counseling intervention only
Zwar 2016	Multiple risk factor study

Methods	Multicenter randomized clinical trial
Participants	Target: 600 people who smoked with a cumulative habit of more than 10 packs of cigarettes per year
Interventions	Intervention: participants will receive usual advice to quit by a general practitioner as well as a 20‐minute personalized visit to provide detailed information about spirometry results Control: participants will receive usual care
Outcomes	Smoking abstinence
Notes	NCT01194596

Methods	Parallel randomized controlled trial with blind evaluation
Participants	942 adults who smoke
Interventions	Intervention: brief advice plus exhaled carbon monoxide measure Control: brief face‐to‐face anti‐smoking advice from the physician during patient consultation
Outcomes	Sustained abstinence (at 6 and 12 months) validated by urine cotinine test
Notes	ISRCTN67499921