Background

Smoking during pregnancy increases the risk of multiple serious adverse infant outcomes including stillbirth, preterm birth, low birth weight, asthma and childhood respiratory infections [1,2,3,4,5]. In 2018, 9.2% of pregnant women in Australia smoked in the first half of pregnancy, and of these, 80% smoked in the second half of pregnancy [6]. While antenatal smoking has declined in recent years, this decline has been unequal and disparities persist: with a persistently high prevalence of smoking among Aboriginal and Torres Strait Islander women (43%), teenage mothers (31%), those living in remote (18%) and very remote (36%) areas, and those of low socio-economic status (17%) [6].

While many pregnant women are highly motivated to quit [7] and are interested in receiving care to achieve this [8,9,10], they often face substantial challenges including a lack of effective support from clinicians [11]. Evidence from systematic reviews indicates that psychosocial interventions to support pregnant women to quit are effective [1, 12]. The missing link, however, is a failure to consistently implement effective cessation care for pregnant women.

Evidence-based Australian guidelines [13] developed by NSW Health recommend routine, repeated smoking cessation support (SCS) for all pregnant women using brief interventions based on the 5As (Ask, Advise, Assess, Assist and Arrange follow-up) and include providing nicotine replacement therapy (NRT) if women are otherwise unable to quit. The 5As are effective and considered best practice, with wide adoption internationally [14,15,16,17]. However, provision of recommended SCS to pregnant women has remained persistently poor both in Australia and elsewhere [10, 18,19,20,21]. A state-wide survey of women’s experiences of maternity care in NSW found that only 46% of women who smoked in pregnancy recalled being told about quitting programmes [22]. Our own research confirmed midwives, obstetricians and managers all reported major gaps in care, particularly in assisting women with cessation strategies and arranging follow-up [23, 24], both of which are crucial to quitting success [25, 26].

There is strong evidence that changing the behaviour of health care providers with persistently poor practice requires the use of a comprehensive evidence-based and theory-driven approach which addresses specific behaviours in specific contexts [12, 27]. One approach, the ‘Behaviour Change Wheel’ (BCW) [28] was developed from 19 frameworks of behaviour change and is based on the COM-B model of behaviour where Capability, Opportunity and Motivation are all required to enact a particular behaviour. The BCW presents a structured approach to developing interventions, with three stages, within which are eight steps—see Fig. 1. We selected the BCW for the development of our intervention as this method starts with the behaviour, allows for a comprehensive analysis of barriers to and enablers of the behaviour and offers a systematic way of selecting intervention content to target barriers and enablers. Furthermore, the clear specification of behaviour change techniques supports subsequent evaluation of implementation fidelity, testing mechanisms of action and future replication.

Fig. 1
figure 1

Behaviour change intervention design method

Full size image

The aim of this study was to use the BCW to develop an intervention to improve implementation of the smoking cessation in pregnancy guidelines [13] in public antenatal care, in NSW, Australia. In Australia, 75% of births occur in public hospitals, and 94% of women have at least five antenatal care visits [6]. While several different models of care are available (e.g. midwifery led, obstetric led or GP shared-care), all women intending to birth in a public hospital should have received at least one antenatal visit within the public health system, and most will have received the majority of their antenatal care in the public system. Thus, improving SCS in these settings has the potential to have a significant impact on the wellbeing of women and their babies. While it would also be beneficial to improve SCS among GPs and obstetricians working in private practice, addressing the barriers in these settings would likely require very different approaches, and thus they are not specifically included in this intervention.

The project was initiated from an ongoing collaboration between academic researchers and a senior Clinical Midwifery Consultant (CMC) (CA) due to evidence of suboptimal implementation of SCS recommendations in Australian maternity services [10, 18]. In NSW, the Ministry of Health determines overall policy and strategy, with clinical care provided by 15 Local Health Districts (LHDs). Throughout the process, we involved stakeholders at multiple levels of the NSW Health system and from several disciplines, which helped strengthen the engagement of these key partners and our understanding of the context in which the intervention would operate. Here, we report the process followed, the intervention developed and the outcomes of a small feasibility study of the intervention.

Methods

The multidisciplinary research team consisted of a public health physician (MP), a social scientist (JL), two researchers with expertise in behaviour change and implementation science (CP&LA), and a senior midwife (CA). All had prior experience in smoking cessation research. At the start of the project, we met with senior officials in NSW Health (covering tobacco control, midwifery and obstetric care) and successfully secured their support for the project and its implementation. The project was guided throughout by an Advisory Group consisting of clinicians (midwives and obstetricians), maternity care managers, policymakers and tobacco control experts, including those involved in the Aboriginal Maternal Infant Health Service and those recommended at the initial meetings with NSW Health. The Advisory Group provided advice and direction to the project, assisted with implementation and ensured relevance and consistency with state policies and initiatives. At several points, the Advisory Group specifically engaged us with relevant state developments to optimise the project outcomes and opportunities for translation into policy and practice. These included contributing to changes to the electronic medical record (EMR) system for maternity services and assisting in the development of new smoking cessation training modules for clinicians.

We followed the stages and steps described in the BCW book [28] and shown above (Fig. 1), then undertook a small study in one site to assess the feasibility of implementing the intervention and acceptability of its components. These steps are detailed below, and the intervention developed (Midwives and Obstetricians Helping Mothers to Quit (MOHMQuit)) is described at the end of the ‘Results’ section at step 8, with subsequent revisions based on the feasibility study described just prior to the ‘Discussion’ section.

Stage 1. Understanding the behaviour

Step 1. Define the problem in behavioural terms

We used Australian smoking cessation guidelines for use in pregnancy [13] and documented evidence-practice gaps identified in the literature [10, 18, 19, 22] to define the problem and the target population.

Step 2. Select target behaviour

As the guidelines recommend provision of the 5As at every antenatal visit, we accepted this as the target behaviour and did not generate any alternatives.

Step 3. Specify the target behaviour

Based on the guidelines and our knowledge of the NSW Health maternity care system, the research team specified who, where, when, how often and with whom the target behaviours should occur, in consultation with the Advisory Group.

Step 4. Identify what needs to change

We undertook two related studies, using the Theoretical Domains Framework (TDF) to inform data collection and analysis [29]. The TDF is a validated comprehensive framework for identifying barriers and enablers to behaviour change and includes 14 theoretical domains that can be mapped to the six components of the COM-B model (physical capability, psychological capability, physical opportunity, social opportunity, reflective motivation and automatic motivation) [28]. We undertook an initial qualitative study with 27 maternity service managers, obstetricians and midwives working in the public health system in NSW, using the TDF to guide the interviews in order to identify the barriers and enablers to provision of SCS [23]. We then conducted an online anonymous survey with midwives working in public antenatal clinics in NSW, with the questionnaire based on the TDF, and refined using findings from the qualitative study [24]. As these studies are already published, we have not presented detailed methods here. Data from both these studies were mapped to the TDF domains and the COM-B components by MP and JL, with review and discussion with the remaining authors. Through this process, we identified the barriers that needed to be addressed to support behaviour change as well as enablers to assist.

Stage 2. Identify intervention options

Step 5. Identify intervention types

Using the published BCW guidance, we mapped the identified TDF domains and COM-B components to the full range of possible intervention types [28]. These potential intervention types were then reviewed by the research team against the APEASE criteria (Affordability, Practicability, Effectiveness and cost effectiveness, Acceptability, Side effects/safety and Equity), considering relevance and feasibility within the context of the NSW public antenatal care system, with reduction in the number retained.

Step 6. Identify policy options

Provision of SCS as a core part of antenatal care is already informed by NSW Health guidelines [13], required by NSW Health policy, and is an identified NSW Health priority with training in smoking cessation available to midwives. Additionally, a new electronic medical record system (eMaternity) was in development and the research team was contributing to enhancements in the section on smoking, as a result of our initial findings, in parallel with progressing our own intervention development. Consequently, no further policy assessment was undertaken.

Stage 3. Identify content and implementation options

Step 7. Identify behaviour change techniques

The intervention types identified during step 5 were mapped to a potential long list of behaviour change techniques (BCTs) using the BCT Taxonomy v1, which includes 93 different BCTs, and is provided in the BCW book [28]. These were discussed by the research team and an initial shortlist developed based on our knowledge of the context in which we were working. We then reviewed the BCTs using the APEASE criteria. Following this initial review, the research team developed a ‘prototype intervention’ which specified the intervention types and BCTs in detail.

The ‘prototype intervention’ was then presented to a stakeholder workshop including members of the Advisory Group and additional policymakers, midwives (including Aboriginal midwives), midwifery managers, educators, smoking cessation specialists and trainers and researchers. The workshop participants (n = 24) worked in groups to review the intervention types and BCTs using the APEASE criteria, recommending inclusion, modification or exclusion. Following this, further discussions were held with senior policymakers in NSW Health, who provided advice on their preferences regarding fine-tuning the intervention. Final decisions on the inclusion of intervention types and BCTs were then made by the research team.

Step 8. Identify the mode of delivery

The research team considered the options for delivery of the intervention within the context of the NSW Health system and developed a detailed implementation plan, including identification of resources required. Discussions with the Advisory Group members and notes from the earlier stakeholder workshop helped inform this process. Development of the resources included input from additional experts and stakeholders, including pregnant and postpartum smokers. The process of resource development is the subject of a forthcoming paper [30].

Feasibility and pilot study

The final stage of development was a feasibility and pilot study at one site to assess operational feasibility and acceptability of the intervention and its components, among midwives and the maternity service leadership team. Assessing feasibility and acceptability prior to a large trial is recommended in order to identify any problems and confirm that the intervention can be delivered as intended [31, 32]. Due to resource constraints, this study did not include the component targeting obstetricians, although this component will be included in the larger trial. The aims of the pilot study were to assess the acceptability of the MOHMQuit intervention to maternity service leaders (managers, educators and those in other leadership roles) and midwives; the feasibility of implementing the leader and midwifery components of the MOHMQuit intervention; the extent to which participants used the various resources and processes developed; participants’ suggestions for changes or enhancements to the MOHMQuit intervention; participants’ perceptions of the value of implementing MOHMQuit more broadly, possible barriers and potential solutions; and the acceptability, face and content validity of data collection tools planned for use in a larger (subsequent) trial; in order to refine the intervention prior to a larger implementation trial. The approach was based on the theoretical framework of acceptability [33].

The study involved running both the leader and midwifery components and provision of all resources. Data collection is summarised in Fig. 2. Maternity service leaders were interviewed prior to the provision of the intervention and 3 months later.

Fig. 2
figure 2

Data collection for the feasibility study

Full size image

Participating midwives completed questionnaires pre-training, post-training and 3 months later. Questionnaires were based on our validated questionnaire previously used in the state-wide survey of midwives to assess provision of SCS and barriers and enablers to doing so [24]. Twenty items in the original questionnaire relating to barriers/enablers were found to form three factors (‘Capability’, ‘Work environment’ (relates to opportunity in the COM-B) and ‘Personal priority’ (relates to motivation in COM-B)), which strongly predicted provision of SCS [24]. To assess the acceptability, face and content validity of the data collection tools, participants were also asked to provide feedback on relevance, acceptability and clarity of the questions; comprehensiveness of coverage of the issues; question order; time taken to complete; and any suggestions for improvement.

Results

Stage 1. Understand the behaviour

Step 1. Define the problem in behavioural terms

The problem identified was the gap between the guideline-recommended SCS and current practice identified in the literature, which was particularly acute regarding provision of assistance and arranging follow-up, rather than asking about smoking status [10, 18, 19, 22,23,24].

Step 2. Select the target behaviour

The target behaviour was delivery of the 5As (Ask, Advise, Assess, Assist and Arrange follow-up) at every antenatal visit as recommended in the guidelines, with the primary target being the Assist and Arrange follow-up elements, as these were most poorly provided [13]. For women who have recently quit, this same process is followed to identify and prevent relapse.

Step 3. Specify the target behaviour

We specified the target behaviour as midwives providing the 5As as specified in the guidelines, at every antenatal clinic appointment, with all pregnant women identified as smokers or who have quit smoking in the last 12 months. While both midwives and obstetricians provide antenatal care to women in the NSW public health system, all women are seen by midwives, and midwives have greater involvement in behavioural concerns than obstetricians, so we initially identified midwives as the target group for our intervention.

Step 4. Identify what needs to change

Barriers to provision of SCS were identified in five of the six components of the COM-B (psychological capability, physical opportunity, social opportunity and reflective and automatic motivation) and 10 of the 14 domains of the TDF, with changes required in each. We also identified enablers which could potentially assist with addressing the barriers, in four of the TDF domains and three of the COM-B components. These key themes and their corresponding COM-B components and TDF domains are shown in Table 1. Further detail is provided in Additional file 1.

Table 1 Barriers to and enablers of the provision of smoking cessation support, TDF domains and COM-B components, potential intervention types and BCTs initially identified
Full size table

Early translation activities

In discussions of these findings with the Advisory Group, we were informed that NSW Health was implementing a new EMR for maternity services—eMaternity. The initial version was in the testing phase but still did not flag smokers and had not addressed the issues identified above. Through members of the Advisory Group, we worked with the developers to ensure some of these issues were addressed including automatic identification of women who smoke in a ‘banner’ on the screen along with any other ‘Considerations’ specific to that woman, making them easy to identify at subsequent visits; fields for recording smoking status and cigarettes/day at any visit; and spaces to record referrals and support provided. This process occurred in parallel with the intervention development.

Stage 2. Identify intervention options

Step 5. Identify intervention types

The results of the research team’s assessment of each of the potential intervention types against the APEASE criteria are shown in Additional file 2. We selected Education, Training, Enablement, Environmental restructuring, Persuasion, Incentivisation and Modelling, as these all met at least half the APEASE criteria, with the remaining criteria unclear and requiring further consultation in step 7. We rejected Coercion and Restriction as neither of these met any of the APEASE criteria and were considered inappropriate in this context. The intervention types selected for each of the identified barriers are shown in column 3 of Table 1.

Step 6. Identify policy options

As explained in the methods, no further policy assessment was undertaken.

Stage 3. Identify content and implementation options

Step 7. Identify behaviour change techniques

Initially, the research team shortlisted 22 different BCTs with several of these addressing multiple barriers. For example, BCT 5.1 Information on health consequences was used to address the barrier ‘Prioritisation of other health issues over smoking’ by considering the impact of smoking relative to other conditions that are well managed (e.g. diabetes), and the barrier ‘Some midwives did not consider referral to Quitline to be effective’ by providing information that referrals to Quitline result in an increase in quit attempts/rates, as well as several other barriers. These are shown in column 4 of Table 1. The results of the review of the intervention types and corresponding BCTs by the stakeholder workshop are shown in Additional file 3. The majority of the BCTs were supported, with minimal change, and were retained. The workshop also recommended inclusion of Aboriginal health workers in the training programme for midwives as they play a key role in antenatal care for Aboriginal women. Following the workshop and discussions with senior NSW Health policymakers, decisions included [1] continue working with NSW Health in further refining eMaternity rather than developing a stand-alone system [2]; work with the NSW Health Education and Training Institute (HETI), which was developing a new online education module to improve clinicians support for smoking cessation in pregnancy—MP was invited to participate in the Expert Group advising HETI, and findings from step 4 above were used in designing the modules—these modules address some of the education and persuasion elements required [3]; inclusion of a component for the obstetric team, as the Ministry of Health emphasised that smoking cessation was core business for all clinicians, not just midwives.

Step 8. Identify the mode of delivery

The MOHMQuit intervention includes multiple components and modes of delivery (see Table 2). It was designed specifically to address gaps in existing programmes and meet the expressed concerns and needs of antenatal clinicians, at both individual and system levels, increasing the likelihood of uptake. All participants will be asked to complete the HETI online modules prior to participating in MOHMQuit, to ensure knowledge of harms of antenatal smoking, the evidence for SCS, use of the 5As, and use of NRT in pregnancy. These modules include the intervention functions Education, Persuasion and Modelling.

Table 2 The MOHMQuit intervention components and modes of delivery
Full size table

The maternity service leadership workshop will occur prior to the clinician-focused components to enhance leadership in supporting clinicians to provide SCS and will be conducted by a senior midwifery trainer. The workshop and accompanying resources include the intervention functions:

  • Education: about the policy and guidelines to use the 5As and how and when to deliver them

  • Training: how to provide leadership in smoking cessation support, how to identify and support champions and how to use the resources provided including running the eMaternity reports and using them for quality improvement (audit and feedback), using the clinic action planning tool and developing local care pathways for smoking cessation

  • Environmental restructuring: providing a template for running eMaternity reports on provision of SCS in their clinic with guidance on how to use the reports for quality improvement (audit and feedback) and providing an action planning tool to use to audit policy and procedures in the clinic then use this for action planning

  • Enablement: the importance of encouraging/supporting staff attendance at training and of encouraging regular discussions of SCS at team meetings, discussing mechanisms they could use to support team discussions and communicate progress (e.g. displays in the staff room) and brainstorming mechanisms and action planning to improve implementation of SCS

  • Persuasion: providing information on the effectiveness of Quitline and the importance of referrals; comparing smoking with diabetes regarding the health consequences and current service provision, to clarify the extent of the gap in assisting smoking cessation; and reframing smoking as an addiction, rather than a lifestyle choice

On completion of the workshop, leaders will be asked to encourage and support clinical staff to attend the relevant workshops (see below), review eMaternity reports monthly and discuss with their team, complete the action planning tool and review annually, develop local care pathways for smokers and execute other actions guided by their action plan.

The workshops for clinicians (midwives, Aboriginal health workers, obstetricians and obstetric trainees) use evidence-based behaviour change techniques shown in Table 1 (e.g. social comparison, modelling, behavioural practice/rehearsal, reframing smoking) and will be jointly provided by a midwifery trainer and a smoking cessation trainer.

These workshops and resources use the following intervention functions:

  • Education: about the policy and guidelines to use the 5As and how and when to deliver them

  • Training: on how to provide the 5As effectively, including motivating change, providing assistance and follow-up, and integrating it into other clinical activities, using demonstration (videos), role play and feedback, and how to use the MOHMQuit resources

  • Environmental restructuring: (in addition to the changes to eMaternity at the state level) guidance on recording smoking information in eMaternity; a range of resources to remind them to provide the 5As and how to do so (training videos, clinic posters, summary guide of the 5As, helpful hints for clinicians, assist and arrange follow-up flip booklet, a reference card for their badge holder, information on NRT); and resources to use with women (a self-help booklet and NRT information)

  • Enablement: brainstorming ways to self-monitor provision of SCS and action planning to address this

  • Persuasion: providing information on the effectiveness of Quitline and the importance of referrals; comparing smoking with diabetes regarding the health consequences and current service provision, to clarify the extent of the gap in assisting smoking cessation; reframing smoking as an addiction, rather than a lifestyle choice; highlighting the excellent communication skills that midwives already have that can be used to address smoking; and patient testimony on the importance of having cessation supported, and how it feels if it is not addressed

  • Incentivisation: praise for providing SCS from leaders and champions

  • Modelling: video demonstrating how to provide SCS without damaging the relationship

A 1-day workshop for midwives and Aboriginal health workers will be provided (run at least twice at each site to allow all midwives and Aboriginal health workers to attend), while a 3-h workshop will be provided for medical staff during regular clinical meetings. Relative to obstetric staff, the workshop for midwives and Aboriginal health workers will cover additional material and include role play, consistent with the greater role they usually play in behaviour change interventions. This structure and approach is consistent with usual practice regarding upskilling in the NSW Health system. Midwifery educators will provide training for new staff using a train-the-trainer model [34]. As an incentive for participation, continuing professional development points will be awarded.

Feasibility study results

The 3-h leader training was attended by three of four invited managers, but none of the three invited clinical educators, who had prior commitments. The full-day midwives’ training was attended by 12 of 15 midwives invited (27 were identified as eligible but 12 were not invited). All (leaders and midwives) had completed the HETI training in preparation for MOHMQuit.

Both the leaders’ training and the midwives’ training were well received, with positive feedback on the day and high scores on the midwives’ surveys (see Table 3). There were some minor suggestions for improvement of the programme at the training, mostly logistical. Additionally, in both the post-training survey and the follow-up survey, midwives requested more ongoing education or follow-up training. Following the training comments from midwives included: “I feel really motivated to do this because I now understand how important it is, I have skills and I’m confident to try it out” and “Amazing day, needs to be made mandatory for all midwives in the LHD”.

Table 3 Mean scores on usefulness of training and resources post-training and at 3-month follow-up
Full size table

Scores on the Capability, Personal Priority and Work Environment factors from the surveys with midwives prior to training, immediately post-training and at 3-month follow-up are shown in Table 4. There was an increase in all scores, most markedly on the Capability factor. While the increase on the Personal Priority factor was less marked, it had a high baseline and exhibited ceiling effects (maximum score of 15).

Table 4 Median scores on barrier and enabler factors from midwife surveys
Full size table

There was only one suggestion for improving the survey instruments (related to response options for model of care), with participants indicating the instruments were easy to understand and complete, and all took less than 10 min.

Despite these positive findings, there were a number of problems identified, both through the interviews with leaders and our own observations and reflections. Many of these were minor logistical issues and are not presented here. Below, we present important concerns related to the design and delivery of the MOHMQuit intervention.

Leadership

While the leaders participated enthusiastically in the training and continued to be enthusiastic about MOHMQuit at the follow-up interviews, they did not undertake many of the leadership roles and activities that were intended. This included repeatedly delaying the midwives’ training and only inviting a limited number of midwives, rather than all midwives providing antenatal care. Following the midwives’ training, they reported not providing an ongoing discussion of SCS or reminding midwives of its importance. However, one of the clinical educators who had missed the leadership training but attended the midwife training had provided some additional training (using MOHMQuit resources) for midwives who had missed the initial training and intended to continue providing training as part of her training schedule. The interviews revealed considerable variation in the understanding of ‘champions’, but none of the leaders had provided any support or guidance to developing them, beyond supporting midwives to attend the training. Those in management roles reported that there were many conflicting priorities and that they were time poor and thus had not been as involved as they would like to have been. However, they had observed the enthusiasm among the midwives and believed the midwives would carry the programme forward, even without the intended leadership.

Use of tools and resources

One of the leadership groups ran the eMaternity report on provision of SCS in their clinic, but had not used it for discussions with team members prior to our follow-up interview. She reported that she had intended to run it monthly and had ideas for how to use the information, but this had not yet eventuated. None of the leadership groups had completed the action planning tool or developed any local care pathways. Although the leaders all indicated in their interviews that they valued the resources for midwives, they had not been proactive in supporting their use. There was some confusion among midwives and also the clinic coordinator about whether the resources had received the appropriate approvals for use from the maternity service executive, and consequently, the resources had not been implemented.

Subsequent revisions to MOHMQuit

The MOHMQuit intervention is being trialled in eight maternity services across NSW. Based on the findings and experience from the feasibility study, a number of additional components are being added prior to the trial, specifically to support leaders in their role (enablement). These include engagement with CMCs in each participating site prior to commencement, to clarify expectations regarding their roles in supporting practice change and implementation of evidence-based care as they have a critical role in strategic development and service delivery and can support the clinic leaders in their role; development of a ‘community of practice’ [35] once sites have received their initial training; greater clarity in the leadership training regarding expectations and next steps, with support to commence the action planning; and assistance for leaders in developing a ‘roadmap’ to plan next steps and deadlines for actions. This final version of the MOHMQuit intervention is specified in Additional file 4 using the TIDieR checklist.

Discussion

This paper describes a systematic, theory-informed process to develop an intervention to improve provision of guideline-recommended SCS to pregnant women who smoke. The MOHMQuit intervention takes a whole of system approach, with elements for all clinicians involved in antenatal care in the public health system (obstetricians, midwives and Aboriginal health workers), and for maternity service leaders. For each clinician group, there is training specifically addressing gaps in skills and confidence, and self-help resources to use with women. At the broader system level, the intervention includes environmental restructuring (e.g. provision of reminders, the ‘Assist and arrange follow-up flip booklet’ and ‘helpful hints for clinicians’, modifications to the EMR) and multiple mechanisms to support leadership for smoking cessation. The intervention was developed using the BCW and TDF [28, 29], and the process was consistent with recommendations for developing complex interventions, including being attentive to future implementation in real-world settings and designing and refining interventions using iterative cycles with stakeholder input throughout the process [36]. MOHMQuit uses multiple intervention types—education, training, enablement, environmental restructuring, persuasion, incentivisation and modelling, with the aim of achieving sustained culture change within the antenatal service.

Although we did not initially intend to undertake change at the system level, the use of the BCW/TDF method resulted in the identification of system-level barriers to provision of consistent SCS, including problems with the EMR, lack of quality self-help resources, no mechanisms for self-monitoring and inadequate leadership. These issues were only made explicit because we specifically asked questions about the TDF domains of behavioural regulation, environmental context and resources, and social influences. The use of the TDF to inform our primary data collection (both the interviews and the survey) ensured that we intentionally covered all TDF domains from the beginning. While the BCW method can be applied to pre-existing literature and data, particularly if resources are limited, this approach risks incomplete coverage of critical issues, and retrospective mapping to the TDF domains and BCW categories may be challenging [37].

The use of the TDF and BCW method with mapping to BCTs was particularly useful in identifying additional options which we may not have otherwise considered, including ‘action planning’, ‘self-monitoring’ and ‘comparative imagining of future outcomes’. In our feasibility study, the process of comparing care pathways and interventions for gestational diabetes, with those for smoking, was particularly powerful in engaging both the leaders and the midwives in recognising gaps in care. The demonstration of the eMaternity reports confirming current poor provision of SCS in their service reinforced this response.

Conducting the feasibility study was critical as it confirmed that the midwifery component was well-targeted, acceptable and feasible, with minimal changes required prior to our larger trial. There were increases on the Capability, Personal Priority and Work Environment factor scores, particularly the Capability factor. The Personal Priority scores were initially high and exhibited ceiling effects. However, the leaders indicated that they had intentionally invited midwives who were known to be more interested in addressing smoking, which may account for the high baseline scores. The request for ongoing/follow-up training from some midwives confirmed the value of the train-the-trainer approach, but also reinforced the need to ensure the clinical midwifery educators engage with the intervention from the start.

The feasibility study also identified the need for further strengthening of the leadership component. Although our initial data collection identified the need for improved leadership, it did not fully explore the barriers to provision of this leadership. For this, the project would have been strengthened by an additional study, specifically exploring these issues. The additional enablement components added following the feasibility study were developed through discussions within the research team (including CA, a senior midwife), with consideration of the context, and usual processes used within the health system. These may be further refined through discussion with CMCs and other senior midwives at sites participating in our larger trial, prior to its commencement.

We had strong engagement with key stakeholders from project inception. The project was initiated following discussions with a senior midwife (CA) who was a core member of the research team throughout. Her role within the health system also facilitated engagement with senior officials in NSW Health, and with other maternity service providers across the state. Through the Advisory Group, meetings with key policymakers at the Ministry of Health, and the stakeholder workshop, we were able to ensure that the intervention developed was suitable to the context, consistent with NSW Health priorities and policies, and acceptable to core target groups. Others have recognised that despite the highly structured approach of the BCW method, it is not a rigid process but requires the team to make subjective and pragmatic decisions throughout [38]. We see this as a strength, as it allows for flexibility and adaptation to the specific context and the particular issue of concern. In our case, the involvement of stakeholders at multiple levels of the system and in every step of the process meant that the researchers were not making decisions in isolation, but were strongly guided by these groups and able to ensure the intervention was suited to the context.

Involvement of stakeholders in several branches of the Ministry of Health (Nursing and Midwifery, Obstetrics, Tobacco Control, and Epidemiology and Evidence), the Cancer Institute NSW, and multiple LHDs also ensured general awareness of the project across the health system and willingness to engage with the research team in multiple ways. This engagement enabled the rapid translation of some critical findings to improve the development of the new EMR and use of the findings in the development of online training modules. These stakeholders have now partnered with us for the larger cluster-randomised stepped-wedge trial of the MOHMQuit intervention, funded by the National Health and Medical Research Council. Despite these positive outcomes from the stakeholder engagement, the timeframe required for the full process made this engagement challenging to maintain, and there was some turnover among organisational representatives.

Sustainability was a critical consideration in the development of the MOHMQuit intervention. For example, in developing an audit and feedback process, rather than having this dependent on the research team, we deliberately created a mechanism for the clinic leaders to undertake the process through a reporting template in eMaternity, with guidance on how to use this for quality improvement. Other examples of designing for sustainability include the service action planning tool, guidance on developing champions, guidance on developing clinical pathways, and the train-the-trainer model, all of which support local leadership in implementing MOHMQuit and enable ongoing refinement. In addition to supporting sustainability, these tools support adaptation of the intervention to suit the local context, an important consideration when designing interventions for broad implementation [39]. It will be critical to assess the uptake of these elements, how they are used, and longer term sustainability of the MOHMQuit intervention as part of the process evaluation of the implementation trial.

Several other groups have used the COM-B or TDF to develop implementation interventions with demonstrated success in improving SCS. The BabyClear intervention implemented in North East England was developed based on findings from a study using the TDF [40] and demonstrated a doubling of referrals to quit smoking services and a significant increase in quitting by the time of delivery [41]. Gould and colleagues used the BCW method to design the Indigenous Counselling and Nicotine (ICAN) QUIT intervention for delivery in Australian Aboriginal Medical Services [42] with good acceptability and feasibility demonstrated in a small pilot study [43]. Informed by the Consolidated Framework for Implementation Research [44], researchers in the Netherlands developed online information sheets and text messages, tailored to different health professional groups (gynaecologists, midwives, paediatricians, practice nurses and respiratory nurses) with demonstrated feasibility, acceptability and reach [45]. While there were significant improvements in both tested and self-reported knowledge among all groups, only paediatricians increased provision of quit advice [45]. However, the low cost of this intervention makes it highly scalable and further research assessing its sustainability is warranted. Similar digital approaches may also be valuable in sustaining the impact of interventions such as MOHMQuit. For all these interventions, the theoretical underpinnings, and detailed process of design need to be well described, to support understanding of what works, for whom and in what circumstances, and thus advance implementation science in this area.

Strengths and weaknesses

Important strengths of this project have been discussed above, including explicit use of theory, strong engagement with key stakeholders from inception, and the feasibility study with subsequent refinements. The clear specification of the intervention development using this process—i.e. identifying barriers to SCS and specifying specific intervention types and behaviour change techniques aimed at addressing them, also supports the design of the evaluation to assess implementation fidelity, testing the intended mechanisms of action and future replication. The most critical weakness was that we did not explicitly explore the barriers and enablers to providing SCS leadership within maternity services. Utilising an additional theoretical framework to assess the organisational context of antenatal care, such as the Model of Diffusion of Innovations in Service Organisations [46], would likely have assisted to address this and other organisational barriers to change. Doing this would have enabled us to include more specifically targeted intervention types and BCTs to address these barriers.

Conclusions

Using the BCW method combined with strong stakeholder engagement from inception has resulted in transparent development of the MOHMQuit intervention, which targets identified barriers to provision of SCS and is developed specifically for the context in which it will be implemented. While others have developed training programmes for antenatal clinicians to improve their provision of SCS, we are not aware of any that have explicitly addressed leadership for SCS nor the broader systemic issues such as the capability of the EMR to support clinicians in this role. The MOHMQuit intervention is being trialled in eight maternity services in NSW, using a pragmatic cluster-randomised stepped-wedge design. The trial includes multiple outcome measures including changes in clinician and leader behaviour; changes in women’s smoking behaviour; a process evaluation to assess acceptability, adoption, appropriateness, feasibility, fidelity, penetration, sustainability and reach and to test the mechanisms of action; and a cost-effectiveness analysis. The stepped-wedge design has the advantage of being able to assess when changes occurred relative to the provision of the intervention, the possible impact of external factors such as policy changes and the extent to which changes are sustained.