Our study identified user needs and expectations of DHI in women trying to conceive, currently pregnant and up to two years post-birth. Overall, participants in the current study had considerable experience interacting with health-related digital tools. They highlighted several needs that were not fulfilled by currently used digital health tools, and indicated the need for DHIs to be personalized and localized. DHI personalization should include two-way interactivity and customized journey while allowing individuals to control privacy trade-offs. DHI localization should include access to local healthcare resources, and local communities and women’s stories. It is important to note that the themes and sub-themes in relation to both concepts may serve as a generalizable framework across different contexts, and can be adapted accordingly with context-relevant information at their foundation.
Behavioral Insights As Actionable Information For Dhi Design
User interviews generate insights that can inform product design of a digital health tool and direct the developers onto the most appropriate intervention method and modality. For an individual to perform a behavior, the Capability-Opportunity-Motivation Behavior COM-B model proposes that an individual requires capability (physical and psychological), opportunity (physical and social) and motivation (reflective and automatic).27 The model presents as a guide to understanding behavior in context and a starting point to develop behavioral targets for intervention design.28 Through understanding what needs to occur for successful behavior change, we are then able to identify specific intervention functions that are likely effective in bringing about that change.27 Examples of intervention functions include education, persuasion, incentivization, coercion, training, restriction, environmental restructuring, modelling and enablement.27 By mapping insights obtained from the interviews onto the COM-B model, Fig. 1 presents a list of factors that can contribute towards an effective DHI for women pre-conception, pregnancy and post-birth. Factors are further discussed below within the context of personalization and localization.
When personalizing DHIs, participants cited surface level personalization (e.g., inclusion of user’s name on the interface) to be an insufficient incentive to encourage continued DHI usage. While participants were interested in tracking their health behaviors, they were only inclined to sustain their tracking behavior if adaptive and timely feedback was provided. This is in line with an emerging intervention design, the just-in-time adaptive intervention (JITAI).29 JITAI aims to provide the right type and amount of support or actionable feedback, triggered by the system, at the precise timing needed to induce the desired change.29 While research into JITAIs for health-related behaviors are in its early stages, a meta-analysis looking at JITAIs for a range of behavior of interests (e.g., healthy diet, mental health, addiction, weight loss) reported up to 0.87 times improvement on health outcomes of JITAI groups over alternative interventions.30 Nevertheless, a recent systematic review reported that many JITAIs for physical activity lack underpinnings of behavioral science theories and evidence to support content.31 Accordingly, DHI innovators should be mindful of incorporating appropriate evidence-based behavior change theories when designing JITAIs.31 Given the potential of wearables and smartphones to continuously track behaviors and acquire ecologically momentary information of individuals, DHIs should aim to pivot from short message systems (SMS) interventions towards interventions that are dynamic and adaptable to changing status and contexts,32 and to the proffered communication channels (e.g., utilizing omnichannel communication).33
Besides timely and adaptive feedback, women expect DHIs to be customized to their current pregnancy phase and circumstances (e.g., prior miscarriages, currently pregnant with twins, second-time mums). In the current study, participants identified the need to have better awareness and information of their own and child’s health to be better empowered in their health-related decision making. Part of an individual’s ability to make informed decisions about their health behaviors is adequate health literacy that is specific to their health context.34 Adequate access and understanding of relevant health information is essential to reducing unhealthy health behaviours.35 As recommended by the 2021 Lancet and Financial Times Commission, by 2030, all governments should implement civic and digital health literacy efforts, including co-creation of digital tools and health narratives to aid health education and combat disinformation.36 In pregnant women, those with adequate health literacy levels tend to make more informed choices with regard to prenatal testing and food selection, and have better understanding of the dangers of smoking during pregnancy.35,37 Nevertheless, health literacy through health education does not automatically lead to empowerment.38,39 For health literacy to translate into empowerment, DHIs should focus on capacity building (i.e., individual’s capacity to use health information effectively) through factoring in individuals’ competences, subjective perceptions of health and health needs.38
Despite keen interest in a personalized DHI, there were mixed responses in relation to providing personal information. While some participants were willing to only provide basic demographic information, other participants expressed the necessity of incorporating medical records (e.g., blood glucose reading, ultrasound scans) into DHIs to allow for dynamic tailoring. This has been similarly observed in various groups of technology users (e.g., smart wearables, smart home speakers) where the personalization-privacy paradox remains a point of discussion.40,41 Personalization has the capacity to increase or decrease user engagement with the technology depending on the perceived privacy risk versus benefits.42 While different types of users (e.g., ambivalent, benefit-oriented) can express varying levels of perceived benefits and privacy concerns, trust remains one of the predictors for privacy attitudes.43 For instance, users’ trust in a website significantly predicts their intention to share data with the website.43 As current participants expressed trust in the government including health-related agencies, and healthcare professionals, DHI initiatives might benefit from collaborative partnerships with government agencies and hospitals, along with notable branding of these partnerships. It should however be noted that trust in government can be highly variable across countries and relating to this authority as a strategy for building trust of the DHI user may be limited to locations such as Singapore.
Participants from the current study are generally tech-savvy and were able to source information from online sources such as medical websites, forums and mobile phone apps. Nevertheless, the majority of participants struggled to find information that are tailored to the local context. Singapore is a multicultural country with three major ethnic groups – Chinese, Malay and Indian, with each group maintaining their own traditional pregnancy or post-birth practices.26,44 Specific to diet, different ethnic groups tend to have different diet practices during pregnancy and post-birth due to traditional cultural beliefs. For instance, Chinese women reported decreased seafood consumption post-birth due to its ‘cold’ and ‘poisonous’ properties, while Malay women reduced beef and eggs consumption due to beliefs that they inhibit general recovery following birth.45 When faced with uncertainty about the effectiveness of traditional practices and the lack of avenues available to obtain verified information, women from the current study were inclined to remain cautious and follow traditional advices from family and friends despite negative or no effects. Accordingly, providing women with access to local resources and health-related information specific to the local context presents as an effective strategy for DHIs with an education component to adopt.
When faced with struggles and setbacks, women in the current study looked to find health-related support and reassurance from other women in similar situations online. This is in line with existing studies that reported a wide range of topics searched on online pregnancy forums, including physical symptoms, struggles with conception, pregnancy complications, baby’s health, nutrition in pregnancy and product recommendations.46,47 In this study, women reported feeling a sense of connectedness and emotionally supported when reading other women’s stories and experiences, especially when the journey was atypical. Patient narrative is a compelling tool that can be effective in communicating health-related information, and influencing attitudes, judgement and behaviors (e.g., promote positive health behaviors, reduce prejudice, increase attitudes towards consumer products).48 Promisingly, interventions with patient narrative components have shown positive effect on patients’ wellbeing and mood.49–51 Given the multi-modal nature of DHI, DHIs have the potential to explore digital storytelling of patient narratives through different forms (e.g., written, short-form videos, online chat). By leveraging digital storytelling as a strategy for knowledge transition and support, DHIs have the potential to enhance accessibility, education and community building for the targeted population.52
Through understanding sources of behavior and what needs to change, we are able to identify key intervention functions, based on the Behavior Change Wheel,27,53 that should be considered when developing DHIs for women pre-conception, pregnancy and post-birth (see Table 4). Education, training, needs satisfaction and modelling are relevant intervention functions that can be useful to target COM-B components identified as relevant for an effective DHI in this population, and guide further development and fine-tuning of behavior change techniques (e.g., should training be geared towards behavior substitution, habit reversal or habit formation?).
Table 4
Examples of recommended intervention functions for women pre-conception, pregnancy and post-birth.
Intervention functions | Definition | Example of intervention function |
Education | Increasing awareness, knowledge or understanding | Providing dos and don’ts for physical activity based on trimester stages to reduce fear of harming pregnancy |
Training | Developing mental or physical skills | Providing local recipe videos to increase ability to cook healthier food for self and child |
Needs satisfaction | Creating experiences that satisfy inherent needs for autonomy, competence, and relatedness | Directing women to safe digital spaces with other women within the same community to encourage connection, learning and support |
Modelling | Providing visible examples for people to imitate or aspire to | Using personal anecdotes involving healthy diet recommendations to improve eating habits |
Note: See DiTommaso (2019) for full list of intervention functions.56 |
[Table 4. Examples of recommended intervention functions for women pre-conception, pregnancy and post-birth about here]
Limitations
As the current study focused on women in Singapore, some of the findings may be specific to the Singapore context. As localization of DHIs to include the local culture and practices emerged as a significant theme, this might differ in Western contexts which most online resources are currently catered towards. Nevertheless, the push towards co-design for DHI may be broadly impactful across a spectrum of contexts, and may serve as an effective catalyst and advocate for realizing relevant solutions to the users. Despite significant differences observed between cultures in intention to adopt technology (e.g., performance expectancy had greater influence on behavior intention of Swiss consumers than Chinese consumers, while social influence asserted more influence on behavior intention of Chinese consumers than Swiss consumers),54 a considerable amount of technology interactions are not tailored to specific economic or cultural groups of interest.55 Accordingly, continued efforts should be made towards designing for inclusivity and equity through consideration of patient population, cultural and linguistic backgrounds, gender, and disabilities.55