We evaluated the development of Baby Buddy with the BCW and its associated taxonomy using a prespecified three-stage process (Figure 1). The research was conducted between November 2017 and December 2018. The research team was given access to all reports, market research, and interview and focus group findings prepared by Best Beginnings to inform the design of Baby Buddy (Multimedia Appendix 1). Guide books containing worksheets were used to deconstruct and retrospectively analyze the development process and the breastfeeding components within the Baby Buddy app [18,22]. Data extraction was performed by one reviewer (LMM) and then checked by a second (AG). They met fortnightly to share and discuss the findings. This was achieved by cross-checking coding, interpretation, and mapping. Any discrepancy was resolved by discussion, and further analysis or content review was undertaken if necessary.

This stage aimed to assess the in-depth understanding of breastfeeding as a target behavior in the development of the app and the context in which it occurs. Barriers and enablers to target behaviors were identified in the provided data (survey, interview, and focus group reports), and then, these were mapped to the COM-B tool [18].

This stage determined the aspects included in Baby Buddy and if they could influence breastfeeding behavior. The middle layer of the BCW was used by the research team to map which “intervention” components could address the barriers and enablers to breastfeeding (restrictions, education, persuasion, incentivization, coercion, training, enablement, modeling, and environmental restructuring) [18]. We then coded these findings using the Theoretical Domains Framework (TDF) (knowledge, cognitive and interpersonal skills, memory, attention and decision-making processes, optimism, beliefs about consequences, intentions, goals, emotions, and social influences) [18]. The outer layer of the BCW was used to map policy categories (eg, policies, guidelines, fiscal measures, service provision, legislation, regulation, communication, and environmental opportunities) [18].

This stage identified evidence of the use of behavioral change techniques (BCTs) within the design of Baby Buddy. We used the APEASE criteria as defined in the BCW (affordability, practicability, effectiveness, affordability, safety, and equity) [18,22]. These steps provided insights into how the content was developed and implemented and to understand the choices made by Best Beginnings as the project progressed. We also described the “active” ingredients that were used in the breastfeeding intervention using the BCTTv1 tool [18,22]. To do this, we viewed and reviewed 39 videos, eight glossary words (“What does that mean?”), and 20 Baby Buddy–generated responses to breastfeeding questions (“Ask me”). We marked the BCTTv1 tool for each technique found in each piece of information reviewed (videos, glossary words, and generated responses).

Baby Buddy maps retrospectively well to the BCW. This may explain why there have been positive results in recent studies [28,29]. Key factors that set the development of this particular pregnancy app apart from many others are the genuine co-design and the use of BCTs most obviously through the included video content.

The use of participatory engagement and co-creation methods in the development of Baby Buddy are two design techniques that have positively influenced decision making, attitudes, and self-efficacy concerning breastfeeding, particularly among those who are socially disadvantaged and younger. We identified several BCTs used in Baby Buddy that could assist women with decision making around breastfeeding (eg, BCT 5.1 Information about health consequences and 9.1 Credible source). BCTs that influence attitudes and self-efficacy were also identified (eg, BCT 5.3 Information about social and environmental consequences and 13.2 Framing/reframing).

This study has several strengths. First, it was performed independent of the development team, using a best practice behavior change framework (BCW) as a guide. Second, content mapping to the BCW was conducted by two independent content experts (a midwife and a neonatologist). These two research team members located in Australia were not employed by Best Beginnings and did not have any financial incentive. Third, retrospective alignment of the BCW tools and BCTs enabled the research team to identify potential opportunities to use BCTs for the future development of Baby Buddy to increase effectiveness. Fourth, our study supports the work of Thomson and Crossland who conducted a mixed methods evaluation using the BCW to identify components that support infant feeding in North West UK [29]. They identified peer support as a facilitator for increasing mothers’ knowledge and building confidence [29]. Finally, we also identified the use of peer-to-peer content as beneficial for breastfeeding as it normalizes breastfeeding and encourages self-determination. Baby Buddy has both of these attributes in the content. Like the work of Crossland et al, our study concluded that Baby Buddy is a supportive parenting resource that could be scaled for impact [28].

A key limitation of this work is the retrospective application of the BCW. Retrospective mapping of the BCW to the app development process was complex and subjective, and relied on Best Beginnings providing multiple development documents. There was a large volume of qualitative reports supplied to us from Best Beginnings that had been collected from many sources and not presented with later academic review in mind.

Using the BCW has inherent coding, interpretation, and application limitations. However, like other studies, we do believe that there is benefit in “retrofitting” interventions to the BCW even though it may have not been used in the design phase [33-35]. Prospective analysis of the app development using the BCW and scientific research would potentially result in a higher quality behavior change intervention tool; however, Baby Buddy was not primarily designed to change behavior and was rather designed as a resource to inform and empower pregnant women.

A secondary limitation is that the evaluation tools we used were designed for text rather than video content. From our assessment, videos within an app appear to be a powerful influence to support behavior change in breastfeeding. The videos take a “show how” approach rather than a didactic “tell to” approach and feature a mixture of experts, support parents, and peer-to-peer voices. However, as the BCW tools were not designed for video discourse analysis specifically, they may miss some of the nuances in video content (eg, gesturing, body language, and tone). Our findings have identified potential areas for improvement in future iterations of the app, and this is useful information given that the app is constantly being improved.

Our work highlights that applying a theoretical framework retrospectively to a mobile health app is possible and results in useful information to understand potential health benefits and to inform future development. To assess the true impact of behavior change frameworks in the design of mobile health apps, high-quality research that measures formative, process, and clinical outcomes for health behaviors is needed. Further development of Baby Buddy as a universal intervention to reduce health inequalities requires robust prospective research that considers effects on the rate and duration of exclusive breastfeeding.