Designing an implementation intervention with the Behaviour Change Wheel for health provider smoking cessation care for Australian Indigenous pregnant women

The BCW is a parsimonious model synthesising many behaviour change theories (see Fig. 1) [20]. At its hub is the COM-B model, which expands to capability (C), opportunity (O) and motivation (M)—these are all needed to produce or change a behaviour (B) [20]. The COM-B model recognises that behaviour is part of an interacting system involving an individual’s or group’s capability (physical and psychological), opportunity (social and physical) and motivation (reflective and automatic). A mid ring on the BCW comprises nine intervention functions, and on the outer ring, there are seven policy-level strategies. The BCW is used to link influences on behaviour, identified by the COM-B, to potential intervention functions and policy categories.

The TDF is an implementation science model comprising 14 domains that can categorise influences on behaviour [21]. These include the following: ‘Knowledge’, ‘Skills’, ‘Social/Professional Role and Identity’, ‘Beliefs about Capabilities’, ‘Optimism’, ‘Beliefs about Consequences’, ‘Reinforcement’, ‘Intentions’, ‘Goals’, ‘Memory, Attention and Decision Processes’, ‘Environmental Context and Resources’, ‘Social Influences’, ‘Emotions’ and ‘Behavioural Regulation’. According to Michie et al., links between the BCW and TDF can be utilised in the analysis and design of targeted interventions [11].

The BCW and TDF have been used previously to analyse health provider and patient or consumer behaviour; for example, Alexander et al. used the TDF and the COM-B to analyse barriers to the use of the Healthy Kids Check, introduced by the Australian Government in 2008 to guide the health screening of pre-school children by GPs and nurses [22]. Gould used the BCW to analyse barriers, enablers and strategies to improve smoking cessation care for pregnant Indigenous women [4].

The BCW is emerging as an important tool to rigorously develop interventions prior to clinical trials. The TDF and BCW have been used to systematically develop complex interventions in other populations and for other behaviours: several targeted health provider behaviour [23‐28]. Sinnott et al. used the BCW to develop an intervention to improve medication management in multimorbidity by GPs, by uniquely developing a collaborative shared decision-making approach between peers [26]. McSharry et al. developed a multilevel intervention to increase delivery of sexual health counselling by cardiac rehabilitation staff [25]. Murphy et al. used the TDF and BCW to develop a capacity-building program to enhance pharmacists’ roles in mental health care [27]. The latter focused on health provider training, using a train-the-trainer approach. Similar to our approach, Murphy et al. highlight the iterative and fluid nature of the developmental process, which is challenging to capture. Relevant to our study, researchers applied the BCW and COM-B model at both the provider and patient levels to develop interventions to encourage long-term hearing aid use [24] and to encourage nurse-led physical activity in patients [21, 28]. More specifically for smoking cessation, for example, the BCW was used to develop a mobile phone App to encourage patients to use UK Stop Smoking Services, prior to a full trial of the intervention [29], and an App to support smoking cessation among pregnant women [30].

In addition to the BCW and TDF, two factors appear to be relevant for developing Indigenous-specific approaches related to tobacco control and smoking cessation. A factor analysis of 47 organisations developing tobacco control messages for Indigenous Australians revealed two important, yet separate factors of ‘cultural understanding’ and ‘rigour’. Aboriginal Medical Services demonstrated strength in their cultural understanding while universities and government organisations demonstrated strength in rigour. A combination of cultural understanding and rigour was applied by few participating organisations. However, these two important factors emerging from the analysis can be seen as an opportunity to bring the best of two worlds together—Indigenous and Western viewpoints [31].

A conceptual model (Fig. 3) depicts the multiple factors that influence smoking behaviours and smoking cessation among Indigenous women [2, 4, 12, 33‐36, 38‐40, 42, 44, 46‐49]. These factors have an impact at the systemic level, provider level and community and individual levels. Although work is going on to attempt to remediate some of these factors (such as the policy-level lobbying to improve access and affordability of oral forms of NRT), this intervention is focused on translating the evidence-based practice for SCC in pregnancy to the Indigenous context.

We identified the behavioural problem as a lack of essential elements in the health providers’ provision of evidence-based SCC for pregnant Indigenous women who smoke, such as prescribing NRT, providing cessation support, involving family members and following up. Health providers are required to provide SCC to pregnant women who smoke. The Australian RACGP guidelines recommend initially using supportive counselling, but if the woman cannot quit, she can be also offered NRT to assist cessation [50].

A systematic review by Okoli et al. [9] and our own systematic review (Gould 2017, unpublished data) revealed that while health providers often ‘Ask’ and ‘Advise’ about smoking cessation to pregnant women, they less commonly ‘Assist’ women to quit smoking. A lack of support from health providers for smoking cessation during pregnancy was cited as a barrier in previous research, with Aboriginal women and other vulnerable groups [41, 51].

The variety of health providers needing to improve access to SCC within the ACCHS is manifold. Depending on the stage of pregnancy, and the services themselves, women will see several different health professionals. We considered all health providers within a service would need to have the same over-arching approach for SCC, so the women would not receive conflicting advice. Thus, a whole-of-service approach was considered.

Several evidence-based practices are recommended for SCC, including the 5A’s (‘Ask’, ‘Advise’, ‘Assess’, ‘Assist’ and ‘Arrange’) [50], the ABC (‘Ask’, ‘Brief advice’, ‘Cessation’) [11] [52] and the AAR (‘Ask’, ‘Advise’, ‘Refer’) [53]. Gould et al. proposed that health providers should take a holistic approach to SCC for pregnant Aboriginal women and recommended using an ABCD approach—A (Ask/assess), B (Brief advice), C (Cessation), D (Discuss psychosocial context) [43].

A survey of 378 GPs and obstetricians’ knowledge, attitudes and practices related to SCC for pregnant women revealed that clinicians seldom perform the 5As of smoking cessation (19.6% always/often) [8]. While more perform the AAR (49.2% always/often), there are few referral options for specialised services for pregnant women, and many are not being referred or followed up. Clinicians fairly reliably (over 75%) asked and advised about smoking cessation, but only a third actually assisted pregnant women to achieve smoking cessation. Cessation support was provided always by 33.6%; NRT was recommended/prescribed always by 11.1% and never by 25.1%. The TDF analysis revealed that lack of optimism, lack of time and lack of resources were the most frequently cited barriers [8].

Similarly, a survey specifically with health providers working with Aboriginal pregnant mothers revealed that only 4.7% recommended NRT to most or all pregnant clients who smoked [40].

Furthermore, Aboriginal women in our qualitative study revealed that they were seldom offered NRT, and when an offer did occur, it was as a ‘one-off’ offering, and the approach and messages were inconsistent [41]. Women reported they were often told to ‘cut down’ their smoking, but rarely supported to quit [41].

A review of this combined evidence led to a decision to primarily target the behaviour of health providers in offering NRT to assist smoking cessation. In other settings, assisting smokers to quit by providing a prescription of NRT was more effective than providing advice (RR 1.68, 95% CI 1.48–1.89 vs. RR 1.24, 95% CI 1.16–1.33) [41, 51]. A parallel target was to support the provision of a holistic culturally competent approach, taking into account the social determinants of health and psychosocial factors related to continued smoking during pregnancy, that is exemplified in the ABCD approach [43], and thus provide resources to target the patients’ needs also.

In specifying the target behaviour, the BCW guidebook recommends consideration of Who, What, When, Where, How often and with Whom. The target behaviour therefore would be for clinicians (Who) in the ACCHS (Where) to proactively offer assistance to every pregnant Indigenous smoker (Whom) to quit smoking (What), initially by counselling, but importantly by offering NRT, if the woman was not able to achieve abstinence in the first 2–3 days of a quit attempt (When). This would ideally be on every occasion that a smoker is seen (How often) until abstinence achieved. Although some guidelines suggest women should try a quit attempt unaided by pharmacotherapy for 2 weeks prior to be offering NRT, we considered this inappropriate in this context. It is notable that in Eades et al.’s RCT, women were asked to do this, and between the second and third visits, there was a dramatic drop-off in attendance from 70 to 30% of the cohort [15]. So as not to lose momentum in the quit attempt, and in the context of wanting women to quit as soon as possible in pregnancy, an expedited use of NRT was favoured [54].

Table 1 summarises the key items that required remediating for health providers, by considering the BCW, the COM-B model and the TDF. Key areas for performance improvement included capability (psychological skills), motivation (optimism) and opportunity (resources/time).

According to the BCW, for this to be successful, clinicians would be required to increase their psychological capability, i.e. knowledge and skills, and reflective motivation, i.e. confidence and optimism, in how to prescribe NRT for a pregnant Indigenous woman who smokes. Lower levels of confidence for NRT prescribing, compared to counselling, were revealed in our national study [8]. Clinicians also reported low optimism that their treatment would be effective [53]. Thus, providers need to build both self-efficacy and response efficacy. Furthermore, as the preferred oral forms of NRT to use in pregnancy are not subsidised in Australia, health providers require improvements in their physical opportunity to provide care, i.e. access to oral NRT supplies, in order to effectively prescribe [47]. Clinicians understanding that these practices are routine in other countries (such as in Australian guidelines and programs and in other countries) may help them improve their social opportunity.

Table 1 also outlines the intervention components developed for health providers according to the analysis. Intervention functions which best met the APEASE (Affordability, Practicability, Effectiveness and Cost-effectiveness, Acceptability, Side effects/safety, Equity) criteria were considered for inclusion [11]. APEASE is a criterion for making context-based decisions on intervention content. Clinicians in our survey indicated several options to improve their smoking cessation care (Gould 2017, unpublished). Training was an option that most of these clinicians agreed on. This is supported by a Cochrane Review that found training of HPs in smoking cessation care increased abstinence in their patients [55]. Therefore, education and training were chosen as the predominant intervention function. However, to address the low optimism, we include persuasion, modelling and enablement in the intervention. Resources were also developed for the women to support health providers in their consultations, and to provide resources for health education, and practical assistance for their patients (described below).

On the policy layer of the wheel, service provision and guidelines were our main policy targets. We wanted to aim our approach to the whole of the ACCHS, in recognition that many health providers and allied health professions have a role in consulting with Indigenous women during their pregnancy. We wanted smoking cessation to be ‘everyone’s business’. Furthermore, we developed a comprehensive treatment manual that gave very practical guidance for the approach, based on the published pragmatic guide and the RACGP guidelines [43, 50, 56].

Table 1 also specifies the behaviour change techniques (BCTs) related to the above analysis and to address each intervention function. A taxonomy of BCTs was developed by Michie et al. (BCTTv1) as a comprehensive resource for intervention development [57]. Each of the 93 consensually agreed, distinct BCTs in the taxonomy are catalogued and described in detail. The BCTs identified for the ICAN QUIT in Pregnancy implementation intervention were those which authors considered as promising to elicit behaviour change in the health providers.

In summary, to implement this smoking cessation intervention optimally, we aimed to improve capability by training clinicians in NRT prescribing, structuring the consultation using a flipchart and prompts and regulating behaviour through audit and feedback (allowing social opportunity). To improve optimism, in the training, we present recent evidence about NRT and positive testimonials from patients and clinicians. We recognised that individual clinicians may not have experienced intervention success but needed to be exposed to that success vicariously through other’s testimonials, so they see that success is possible and worthwhile persisting for (response efficacy).

To accommodate challenges around time for health providers, the large geographical area of Australia and the limited resources of the research, we decided to trial training via interactive webinar. Webinar was chosen as a mode of delivery with the potential to increase the reach of health provider training to urban, rural and remote locations in Australia and improve the potential scalability. Although webinar is now a very common method of distance training, there is little research in the peer-reviewed literature about the efficacy or effectiveness of this method. However, we considered webinars to be affordable, cost-effective, safe, likely to be acceptable and highly equitable. They give the potential for face-to-face contact.

Training of short duration (40 to 120 min) has been shown to increase smoking session outcomes in patient outcomes in a Cochrane Review [55]. Thus, three 1-h webinar sessions were planned to bring the opportunity of training to services and accommodate time and location constraints, which may otherwise limit attendance by the health providers. Webinar sessions were to be interactive and include PowerPoint presentations, short videos and group discussions. These went on to be trialled in the pilot study in six services in three states [58].