Background

Tobacco use and exposure results in more than eight million deaths worldwide each year [1], prompting an urgent need to implement interventions to promote smoking cessation. There are currently three pharmacotherapies approved for smoking cessation by the US Food and Drug Administration (FDA): varenicline, bupropion, and nicotine replacement therapy (NRT) [2]. A Cochrane systematic review reported that, compared to bupropion or NRT, varenicline is the most effective pharmacotherapy for maintaining long-term smoking abstinence (at six months or more) [3]. A high-affinity partial agonist at the α4β2 nicotinic acetylcholine receptor, varenicline decreases the rewarding effects of tobacco through its dual effects as an agonist by binding to the receptor to reduce craving and as an antagonist by competing with nicotine for the receptor [2, 4]. Despite varenicline being superior to other pharmacotherapy in the treatment of tobacco dependence, low adherence to varenicline is a significant obstacle to the success of this smoking cessation treatment [5]. Meta-analyses have demonstrated the association between varenicline and adverse effects such as nausea, constipation, flatulence [6], sleeping disorders, insomnia, abnormal dreams, and fatigue [7]. In a retrospective cohort study examining varenicline adherence, 55% of the study participants never began their 12-week treatment, 20% began but failed to complete their treatment, and only 25% of the participants adhered to and completed their treatment [5].

Studies have shown that providing behavioural supports and tailored interventions can increase adherence to smoking cessation medications [8].These studies have a large variability in the strengths of effects [9,10,11,12] which may be accounted for by the active ingredients in the behavioural supports the intervention offered. In addition, there is no review examining the behaviour change theory that could guide the design of the intervention targeting varenicline adherence. This is a significant shortcoming given that there is growing evidence, including the UK Medical Research Council's (MRC) framework for complex interventions [13] supporting the use of theory in complex interventions. Theory holds the potential to enhance researchers' comprehension of the behavior change process and provide guidance in the development and refinement of interventions [14]. For instance, theory can help identify theoretical constructs to target within the intervention (e.g. 'optimism.'). Therefore, before designing an intervention to help people adhere to their varenicline treatment, it is essential to conduct a review exploring modifiable determinants that influence varenicline adherence, grounded in a theoretical framework. The Behaviour Change Technique Taxonomy version 1 (BCTTv1) provides a practical taxonomy to describe the active content of an intervention [15]. Behaviour change techniques (BCTs) can be mapped to the Theoretical Domains Framework (TDF) [16], a framework that integrates 33 theories and 128 constructs into a single framework that contains 14 domains [17]. The TDF, in turn, can be mapped to a well-established model of behaviour change: the Capability, Opportunity, and Motivation Model of Behaviour (COM-B). COM-B suggests that behaviour change results from an interaction between people’s capability, motivation, and opportunities for the behaviour [18].

The aim of this rapid review is twofold: 1) to identify the modifiable barriers and facilitators to varenicline adherence in people using varenicline for smoking cessation, and 2) to identify the behaviour change techniques associated with helping people adhere to their varenicline treatment.

The findings from this review will inform the design of a theory-based healthbot planned to improve varenicline adherence in people undergoing smoking cessation treatment.

Methods

We chose to conduct a rapid review since it is a timely, cost-effective and efficient way to gather high-quality evidence to inform health program decisions [19]. The rapid review was conducted in accordance with the Cochrane rapid review methods recommendations [20] and it is reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (see Additional file 1) [21]. The study was registered with PROSPERO (# CRD42022321838).

Eligibility criteria

The research question was developed using the PICO model.

  • Population: The population of interest were individuals using varenicline for smoking cessation.

  • Intervention: Studies were included if varenicline was used as an intervention for smoking cessation. We included studies using multiple smoking cessation medications, as long as they reported factors associated with only varenicline users separately.

  • Comparator: In studies with a comparator group, the comparator was either a placebo, an active control group, or no intervention

  • Outcome: The outcome of interest was reported modifiable factors associated with adherence to varenicline.

Exclusion criteria

  1. 1.

    Publications such as commentaries, abstracts, conference papers, reviews, editorial letters, protocols, book chapters, thesis/dissertations, case reports, and case series.

  2. 2.

    Studies that did not separately report barriers and/or facilitators associated directly with varenicline adherence.

  3. 3.

    Studies in which varenicline was not administered for smoking cessation.

  4. 4.

    Non-English language articles.

  5. 5.

    Non-peer reviewed articles.

Information sources and search strategy

The search strategy was developed with a health sciences librarian (TR), who conducted all searches. The strategy was tested and finalized in MEDLINE (Ovid), then translated and run in the following bibliographic databases: MEDLINE, Embase, Cumulative Index to Nursing & Allied Health Literature (CINAHL), APA PsycInfo, and Cochrane Central Register of Controlled Trials (CENTRAL).

The search strategy was designed to identify the overlap between three concepts: tobacco smoking, varenicline, and treatment adherence (see Additional file 1). The smoking concept was kept broad (e.g. “smoking”, “nicotine”, “tobacco” and relevant subject headings) and functioned only to omit alternative uses of varenicline (i.e. treatment of dry eye syndrome) [22]. The varenicline concept included generic, and brand names (“varenicline”, “Chantix”, “Champix”) searched in the major record fields.

The treatment adherence concept used database-specific subject headings, natural language keywords, and advanced search operators such as truncation and adjacency operators to balance specificity and sensitivity. Variations of search terms such as “retention”, “dropout”, and “compliance” were searched in the title, subject heading, and keyword fields and were linked with “therapy” or “treatment” or “program” using an adjacency operator to search the abstract field. Terms such as “barrier” and “facilitator” were searched in the title, subject heading, and keyword fields and were linked with treatment or retention terms using an adjacency operator to search the abstract field.

The terms and concepts were combined using Boolean operators. Non-human animal studies were excluded [23], as were the following publication types when possible: book chapters, dissertations, conference abstracts, editorials, and letters. Year limit applied was 2006 to the date of the search (May 6, 2022) to reflect the FDA’s approval year of varenicline [24]. The core MEDLINE search strategy can be found in Additional file 2.

The studies located by the research librarian were imported into the reference manager, EndNote [25], and then uploaded into the systematic review software, Covidence [26]. Articles with duplicates were tagged and removed in Covidence [26].

Study selection process

All reviewers conducted a pilot exercise on Covidence to calibrate and evaluate the review forms used in the title and abstract screening, full-text screening, data extraction, and quality assessment. For the pilot screening, all reviewers conducted title and abstract screening on 39 studies and conducted full-text screening on five randomly selected studies that were included in the title and abstract screening stage [20].

Two reviewers independently screened 201 studies for the title and abstract screening, which included resolved conflicts. Afterwards, one reviewer screened the remaining abstracts while a second reviewer screened the abstracts deemed irrelevant by the first reviewer. Given that modifiable factors associated with varenicline adherence could not always be determined in the title and abstract, only the “yes” and “no” options on Covidence were used in the title and abstract screening, where “yes” was selected if the abstract was ambiguous or suggested the reporting of barriers and/or facilitators to varenicline adherence. Studies with missing abstracts also received a vote for “yes” and eligibility was determined in full-text screening. Two reviewers were required for full-text screening, where one reviewer screened all the included full-text articles (MW) while a second reviewer screened the full-text articles excluded by the first reviewer. Conflicts were resolved by a third reviewer or by consensus [20].

Data extraction

Utilizing the revised data extraction form from the pilot (see Additional file 3), data extraction was performed by two reviewers. One reviewer extracted data using the data extraction form and a second reviewer verified the accuracy and completeness of the data extracted by the first reviewer. Conflicts were resolved by a third reviewer or by consensus [20]. Missing data were obtained by contacting the corresponding authors of the included studies. Extracted data included:

  • 1. Barriers and facilitators associated with varenicline adherence.

  • 2. “Active ingredients” employed by the varenicline adherence intervention, defined as the components of a behaviour intervention that are needed for it to work and are observable, replicable and irreducible [25],

  • 3. Study information including: sample size, location of intervention, study design, theories used to design the intervention, delivery of intervention, method of smoking cessation (e.g., abrupt cessation, gradual cessation via reduction), and type of tobacco product used.

  • 4. Demographic information: gender proportion, target population of intervention, age, race, and any additional demographic information reported.

  • 5. Information regarding varenicline adherence including: definition of varenicline adherence, adherence outcome measures (e.g., self-report, pill count), and degree of non-adherence (e.g. discontinuation, reduction). For studies in which adherence to varenicline was not the primary outcome, adherence was defined as adherence to the varenicline treatment. Participants who failed to adhere to their varenicline treatment (e.g., discontinued or stopped taking varenicline but were still in the study) were considered non-adherent.

Barriers and facilitators associated with varenicline adherence were extracted and defined according to the TDF, version 2 [27]. For studies that aimed at improving varenicline adherence, we used BCTTv1 [15] to extract data on the components of the intervention (active ingredients).

Methodological quality assessment

We used the Joanna Briggs Institute’s (JBI) Critical Appraisal Tools [28] to assess the quality of randomized controlled trials (RCTs), quasi-experimental, analytical cross-sectional, case–control, cohort, and qualitative studies, and the Mixed Methods Appraisal Tool (MMAT) version 2018 [29] to assess the quality of mixed methods studies. For studies using the JBI Critical Appraisal Tool, an overall score was calculated based on the percentage of “Yes” answered, and questions were excluded from the overall score if “Not applicable” was answered. Studies with an overall score of 70% and above were deemed low risk of bias, studies with a score between 40 and 70% were deemed moderate risk of bias, and studies with a score of 40% and below were deemed high risk of bias [30]. Secondary and pooled analyses were assessed using the RCT checklist, and reference was made to the parent study. Since the use of an overall score to determine the quality of a study is not recommended for the MMAT, a detailed presentation of the quality assessment was provided for all included studies using the methodological quality criteria from the MMAT to determine whether it was of high, moderate, or low risk of bias [31].

The quality assessment was performed by two reviewers. One reviewer rated all included studies using the quality assessment form and a second reviewer verified the appraisal made by the first reviewer [20]. Conflicts were resolved by a third reviewer or by consensus.

Data synthesis

We used a narrative synthesis of the included studies [20] to summarize the barriers and facilitators to varenicline adherence and the active ingredients in interventions that aim to help people adhere to varenicline. Barriers and facilitators were coded based on the 14 domains of the TDF (version 2). Active ingredients were coded based on the 16 groups of the BCTTv1. All reviewers were trained to code using the TDF and BCTTv1 (http://www.bct-taxonomy.com/). Discrepancies in coding were resolved by consensus or by an expert in TDF and BCTs (NM).

In order to understand which BCTs helped with adherence, we categorized the interventions into three simple categories: ‘effective’, ‘mixed results’ or ‘ineffective’. An intervention was categorized as ‘effective’ when improvements to medication adherence were reported to be statistically significant compared to the control group or baseline measures. Interventions were categorized as having ‘mixed results’ when the BCTs increased the participants’ knowledge, skills, or motivation but showed no sign of improving medication adherence. Interventions were categorized as ‘ineffective’ when the intervention did not significantly improve medication adherence compared to the control group or baseline measure.

Studies with low and moderate risk of bias were used to examine barriers and facilitators to varenicline adherence. In contrast, studies with high risk of bias were only used to confirm the patterns identified.

Results

A total of 1,221 titles were identified through the database searches; 61 met the eligibility criteria (see Fig. 1). Of these 61 studies, nine reported BCTs used to help participants adhere to varenicline.

Fig. 1
figure 1

PRISMA flow diagram detailing the identification, screening, and inclusion of studies in the rapid review

Full size image

Study characteristics

Most of the studies included in this review consisted of RCTs (n = 38) [32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69], followed by cohort (n = 17) [70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86], cross-sectional (n = 3) [87,88,89], quasi-experimental (n = 3) [90,91,92], and qualitative (n = 1) [93] studies, and one study (n = 1) [53] was a mediation analysis that examined an observational study and RCT. The majority of studies had a low to moderate risk of bias [27 studies were of low risk of bias [37,38,39, 42, 44, 47,48,49,50,51,52,53,54,55,56, 61, 65, 71,72,73, 75, 83, 84, 86, 88, 92, 93]; 28 studies were of moderate risk [32, 33, 35, 40, 41, 43, 45, 46, 59, 60, 62,63,64, 66,67,68,69,70, 77,78,79,80,81,82, 85, 87, 89, 90]; and six studies were of high risk of bias [36, 57, 58, 74, 76, 91]]. Studies were conducted in a multitude of countries across all continents except Antarctica.

All studies focused on adult populations, and most focused on the general public (n = 43) [32, 33, 35, 37, 39,40,41,42,43,44,45, 48,49,50,51,52,53, 55, 59, 60, 62, 63, 65, 67, 69,70,71,72,73,74, 76, 77, 79,80,81, 83, 85,86,87,88, 90, 91, 93]. A few studies investigated specific patient populations: cancer (n = 3) [34, 36, 82]; chronic obstructive pulmonary disease (n = 3) [61, 75, 78]; human immunodeficiency viruses (n = 3) [54, 58, 66]; psychiatric conditions (n = 2) [46, 84]; and people undergoing substance use disorder treatment, including methadone treatment (n = 2) [47, 56]. There was a fairly even gender split among participants in the included studies, although 20 studies reported 30% or fewer female participants [71, 72, 38, 39, 77,78,79, 46, 48, 80, 81, 89, 54, 84, 55, 58, 92, 65, 66, 86]. Table 1 provides a summary of the included studies.

Table 1 Descriptive summary of included studies
Full size table

Barriers and facilitators – by Theoretical Domains

Most studies included in this review reported barriers as opposed to facilitators. Of the 61 studies, 51 studies [32, 33, 36,37,38,39,40, 42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57, 59,60,61,62,63,64,65, 67,68,69, 71, 73,74,75,76,77,78, 80,81,82,83, 86,87,88,89,90,91,92] only mentioned barriers, while four studies [41, 58, 66, 85] only mentioned facilitators, and six studies mentioned both barriers and facilitators [35, 37, 70, 79, 84, 93]. Definitions of the theoretical domains according to Atkins, et al. [94] can be found in Table 2.

Table 2 Barriers and facilitators to varenicline adherence according to the Theoretical Domains framework
Full size table

‘Belief about consequences’ was the most common barrier to varenicline adherence, reported by 53 of the included studies. Moreover, 52 of these studies reported side effects as a barrier to adherence (27 of the studies were of low/moderate risk of bias). The most frequently reported adverse effects contributing to varenicline non-adherence were nausea (n = 23; 20 low/moderate risk of bias) [32, 40, 45,46,47, 49,50,51, 53, 57, 60, 62, 63, 65, 73, 75,76,77, 81, 82, 86, 89, 91]; insomnia/sleep problems (n = 12; 8 low/moderate risk of bias) [36, 39, 47, 54, 56, 57, 63, 65, 73, 76, 86, 91]; headache (n = 8; 7 low/moderate risk of bias) [40, 47, 63, 65, 73, 76, 81, 89]; depression (n = 7; 4 low/moderate risk of bias) [32, 36, 43, 56, 62, 76, 91]; vomiting (n = 6; 5 low/moderate risk of bias) [36, 43, 46, 47, 65, 77]; and abnormal dreaming (n = 5; 4 low/moderate risk of bias) [57, 73, 82, 86, 89].

Another frequently cited domain mentioned as a barrier to varenicline adherence was optimism/pessimism, which was reported in 17 studies (n = 17; 15 low/moderate risk of bias) [33, 35, 43, 50, 51, 55, 57, 73, 74, 79,80,81, 84, 86,87,88,89]. In these studies, participants discontinued their varenicline treatments as their confidence that varenicline would help them quit smoking diminished.

Ten studies (n = 10; 9 low/moderate risk of bias) [35, 43, 68, 73, 74, 78, 79, 84, 87, 88] mentioned beliefs about capabilities as a determinant of varenicline adherence. These studies reported perceived competence and/or low willpower as contributors to varenicline discontinuation.

Nine studies mentioned environmental context and resources (n = 9; 8 low/moderate risk of bias) [43, 52, 68, 74, 75, 78,79,80, 90] as a barrier to varenicline adherence. The cost of the medication and lack of access to pharmacies were the most frequent barriers coded under this domain.

Other less frequently reported barriers to varenicline adherence include: behavioural regulation (n = 3; 2 moderate risk of bias) [74, 79, 81]; memory, attention, and decision processes (n = 2; 2 low/moderate risk of bias) [68, 92]; goals (n = 1; 1 low risk of bias) [84]; and intentions (n = 1; 1 moderate risk of bias) [78].

Facilitators that promoted adhering to varenicline include: social influences (n = 4; 4 low/moderate risk of bias) [37, 41, 70, 79]; knowledge (n = 3; 2 low/moderate risk of bias) [37, 41, 58]; beliefs about capabilities (n = 3; 2 moderate risk of bias) [35, 58, 66]; environmental context and resources (n = 3; 3 low/moderate risk of bias) [79, 84, 85]; and behavioural regulation (n = 1; 1 low risk of bias) [93].

There were few instances where modifiable determinants related to ‘goals’ and ‘intentions’ were reported. When they were reported it was always as barriers. On the other hand, “knowledge and “social influence” were only mentioned as facilitators (Figs. 2 and 3). Table 2 provides a summary of the barriers and facilitators to varenicline adherence according to the Theoretical Domains Framework.

Fig. 2
figure 2

Barriers to varenicline adherence

Full size image
Fig. 3
figure 3

Facilitators to varenicline adherence

Full size image

Behaviour change techniques

Only nine studies included in this review reported on behaviour change techniques associated with varenicline adherence [37, 41, 43, 56, 58, 66, 89, 92, 93]. Among these studies, two studies reported statistical significance in regards to improving varenicline adherence [43, 58], two studies were not statistically significant [37, 66], and five studies did not report statistical significance [41, 56, 66, 89, 93].

The most common behaviour change techniques (BCTs) implemented in the studies for improving varenicline adherence were social support (n = 6) [37, 41, 43, 56, 66, 93]; feedback and monitoring (n = 5) [41, 43, 56, 66, 93]; and shaping knowledge (n = 4) [37, 41, 58, 89]. Other BCTs that were mentioned included goals and planning (n = 2) [56, 89]; regulation (n = 2) [56, 89]; and self-belief (n = 1) [58]. Given that so few studies reported statistical significance, we did not identify any trends indicating which BCTs were promising. Table 3 provides a summary of the BCTs used in these studies.

Table 3 Behaviour change techniques used to improve varenicline adherence
Full size table

Discussion

The goals for this rapid review were to identify: (1) the facilitators and barriers to adhering to varenicline; and (2) the active ingredients utilized in the intervention for varenicline adherence. The results of the review will be used by the authors to help design a healthbot aimed at helping people adhere to their varenicline regimen.

The current review identified 61 studies that identified barriers and/or facilitators to varenicline adherence. Of the 61 studies, nine explicitly mentioned behaviour change techniques used to help with varenicline adherence. Similar to what other evidence syntheses have found on medication adherence [95], our review found a greater emphasis on the barriers than on the facilitators.

By using the TDF framework to extract and analyze the data, we saw that there are eight domains that act as barriers to varenicline adherence (behavioural regulation, memory, goals, intentions, beliefs about capabilities, beliefs about consequences, optimism/pessimism, and environmental context) and five domains that act as facilitators (knowledge, behavioural regulation, beliefs about capabilities, social influences, and environmental context). In this review, side effects were coded under the domains of ‘beliefs about consequences’ since we assumed it is not the side effects per se that influence medication adherence but more of an individual’s acceptance that the medication may cause some unpleasant side effects (such as nausea, sleep disturbances). Under this assumption, the patient’s initial belief that varenicline will lead to side effects may discourage them from beginning their varenicline treatment.

Our results align with other studies, which identified side effects, especially nausea, as a major determinant to varenicline adherence [34, 52]. Additionally, our results are in line with a review investigating factors influencing adherence to Nicotine Replacement Therapy among individuals aiming to quit smoking [96]. This review also identified beliefs about consequences, behavioural regulation, memory, intentions, beliefs about capabilities and environmental context as a significant determinant to medication adherence [96]. Researchers studying different populations (i.e. people with bio-polar disorder, diabetes), have also found beliefs about consequences as a significant determinant to medication adherence [95, 97].

With the exception of the role of optimism/pessimism as a determinant, our findings are similar to other studies using the TDF to understand the determinant of medication adherence [95, 98,99,100]. In our review, we identified 17 studies reporting pessimism as a barrier to varenicline adherence, which contrasts with the results of other reviews which did not identify pessimism/optimism as a determinant to medication adherence [95, 101]. While it might be a unique case that pessimism is a determinant for varenicline adherence and not to other medications, it is more likely that the difference is due to decisions on how to code certain determinants. Several researchers (who did not use the TDF in their studies) have identified pessimism as an important domain for medication adherence [102,103,104].

Similar to what other researchers have shown, we found that providing social support, feedback and monitoring, and shaping knowledge were the most common BCTs used to help people adhere to their medication regimen [105,106,107].

Strengths and limitations

Our search strategy was comprehensive and was developed with the help of an experienced health sciences research librarian. The included studies were conducted in several countries, and there was representation from all continents, with the exception of Antarctica. In addition, the included studies used a variety of study designs, allowing for a comprehensive list of modifiable determinants of varenicline adherence to be identified. However, due to the nature of rapid reviews, some relevant studies may not have been captured (e.g., exclusion of non-English publications, proceedings and relevant information in the gray literature).

Utilizing the TDF to organize our data, we were able to focus on modifiable determinants and, at the same time, map them to a well-defined theory of behaviour change. However, as mentioned earlier, there were some levels of subjectivity in coding for a few determinants.

Given that very few studies reported BCTs, and of those that did, most did not report the statistical significance of their results, we were unable to examine trends on what BCTs are promising when targeting varenicline adherence.

Conclusion

Using the TDF framework, our analysis revealed eight domains as barriers (behavioral regulation, memory, goals, intentions, beliefs about capabilities, beliefs about consequences, optimism/pessimism, and environmental context) and five domains as facilitators (knowledge, behavioral regulation, beliefs about capabilities, social influences, and environmental context) to varenicline adherence. The insights into these barriers and facilitators provide valuable guidance for healthcare providers and decision-makers in shaping the design and delivery of smoking cessation services incorporating varenicline. Future work will explore how a healthbot [108, 109] could address the barriers identified in this review.