We included papers discussing digital interventions that satisfied all the following criteria: (1) full text in English, (2) targeted healthy human subjects regardless of breastfeeding role, and (3) provided direct or partial breastfeeding support.

We also included publications that discussed a system from different perspectives, for example, from experience-centric and clinical aspects. Work-in-progress papers were also included if their fully implemented system was not available.

We excluded papers that met any of the following criteria: (1) contained only abstract, (2) targeted subjects with special medical conditions, for example, preterm infants or severely sick mothers, and (3) reviews, books, book chapters, or reports or papers from scientific magazines.

If a paper of a fully implemented system is available, the work-in-progress papers of that system will be excluded.

In the full-text assessment phase, we eliminated 26 publications that did not fit our criteria (see Figure 1). Finally, we included 35 eligible papers that described 30 digital interventions. Of the papers published, 31 were in medical or health informatics journals. The other 4 papers were presented in human-computer interaction and computer science conferences.

In total, 11 of the 30 systems aimed to provide breastfeeding education. Here, we separate educational interventions for mothers and training tools for health care practitioners.

A total of 9 papers [25,33-40] studied 8 educational interventions for mothers. The majority of these interventions focused on digitizing existing knowledge into generic learning modules with multimedia to aid explanation [25,34,35,38-40]. Geoghegan–Morphet et al [25] offered a Web-based forum in addition to educational resources. Besides these generic learning programs, some educational interventions were tailored. For instance, Abbass et al [33] involved an indigenous community to craft culturally relevant educational resources. Similarly, Joshi et al [35,39] customized educational content to suit Hispanic mothers.

Other modified generic systems for mothers emphasized learning via interactive exploration. For example, an interactive agent [36] guided mothers to explore different aspects of the breastfeeding process through simulated conversations. Grassley et al [37] proposed a quest-based game for mothers to build up their breastfeeding knowledge by completing playful quests. The game engaged users in various Web-based learning activities, such as reading and watching breastfeeding-related multimedia contents.

We identified 6 papers [23,41,42,44-46] that discussed 4 training tools for health care professionals. These systems aimed to improve breastfeeding knowledge and support skills. A total of 3 papers [42,45,46] evaluated Breastfeeding Basics [43], a Web intervention that provided modular breastfeeding educational resources and training materials for health care practitioners. A total of 2 papers [23,41] focused on designing course contents to be published on off-the-shelf e-learning platforms, such as Moodle and Blackboard Learn. Meanwhile, 1 study [44] made use of a website with 2 separated forums: one for pediatric residents and the other for breastfeeding mothers. This intervention allowed the pediatric residents to learn from their peers on one forum and apply their new skills when supporting mothers on the other forum.

In the second cluster, we identified 8 publications [24,47-53] that described systems that encouraged breastfeeding. The majority [24,48,49,52,53] of the systems utilized text messages to send out personalized breastfeeding tips and encouragement messages and sample breastfeeding experiences via short text message (SMS) responses. The content of the text messages was based on focus group discussions [53], WHO guidelines, expert inputs, and literature studies [49]. A total of 2 interventions [48,52] did not specify the source of message content and 1 system [24] used the Health Belief Model [66] to frame promotion strategies. Maslowsky et al [51] proposed a two-fold intervention that required nurses to call mothers 48 hours after hospital discharge. The idea is to deliver a maternal education session and a follow-up phone call to sample experience and provide support if needed. To keep users engaged with the intervention, Wardle et al [47] encouraged mothers to donate breastmilk using positive reinforcement techniques in their mobile app. The Growing Healthy Program [50] encouraged breastfeeding with weekly personalized motivational messages and on-demand breastfeeding resources aimed at parents with low socioeconomic status.

The third cluster comprises 7 papers [54-59,61] discussing 6 interventions that facilitate communication between peers and professionals. Of the systems, 3 [55-57] provided tele-lactation consultations via video conference for mothers with limited access to maternal care. Giglia et al [59] designed a multichannel support intervention through focus group; the intervention provides lactation supports through Web-based resources, email, and video call, all at the same time. For mothers who were not satisfied with their current caregivers, Thomas and Shaikh [58] described how mothers sought breastfeeding supports on the Web, using various information and communication tools. Besides interventions for mothers, White et al [54,61,67] used the social cognitive theory to design a mobile app that fostered peer support among expecting and new fathers by engaging them in discussions. This mobile app grouped fathers based on their baby’s age and maternal stage of his partner.

In the fourth cluster, we found 3 mobile apps and a Web intervention that practically supported breastfeeding. FeedFinder [62], MoomMae [63,65], and MilkTrack [64] are mobile apps that allow mothers to find, rate, and review suitable breastfeeding locations in public spaces. Besides breastfeeding location crowdsourcing, other features were incorporated as well, for example, MilkTrack [64] provides practical breastfeeding resources and a milk donation platform, whereas MoomMae [63,65] provides a personal breastfeeding diary. Ahmed et al [60], in contrary, only provided a Web breastfeeding diary. The diary allows health care practitioners to gain insights into mothers’ breastfeeding experience and personalize their support.

On a general level, this shows that ICT to support breastfeeding addresses a broad range of goals; however, the strongest categories that emerged throughout our analysis were systems to provide breastfeeding education and promote breastfeeding.

Client communication systems were used in 18 digital interventions [24,25,33,34,38,44,47-54,59,61-65] to communicate breastfeeding information to targeted clients (parents), provide on-demand information services to clients, and facilitate client-to-client communication. These systems leveraged 5 technology platforms to reach their intended audiences.

SMS appeared in 6 interventions [24,48,49,51-53] that aimed to deliver breastfeeding information and encouraging messages to mothers. Most of them [24,48,49,52,53] were inspired by the trend of leveraging SMSs to promote healthy behavior. Affordability and high availability also contributed to the popularity of this technology [48,49,52], particularly in developing countries, as 5 SMS-based breastfeeding interventions [24,49,51-53] were studied in Myanmar, China, Ecuador, and African countries, whereas only 1 [48] was introduced in Australia. The degree of user interactions within these SMS-based interventions is rather limited: 2 systems [24,48] required users to reply with a predefined code or a syntax, whereas the other 4 [49,51-53] invited free text response, but manual human processing was required to extract the content.

Phone calls, which operate on the same infrastructure as SMS, were used together with SMS in 2 interventions [51,52] to compensate for the lack of in-depth communication.

A total of 2 interventions [50,59] used email to initiate conversations with peers [59] and deliver notifications and weekly messages in a mobile health intervention [50] that also provided personalized infant-feeding information through a mobile app.

A total of 6 mobile apps [47,50,54,61-65] provide on-demand information service and client-to-client communication via platforms for online discussion and information sharing. Mobile app development gained popularity among intervention designers following the ubiquity of smartphones [47,50,54,61-65]. Mobile apps can be used to establish social connectivity for parents to connect to peers and foster information sharing, for instance, review public breastfeeding spaces [62,63,65], facilitate milk donation logistic [64], and share parenting experiences [50,54,61]. In addition to social connectivity, breastfeeding information can also be provided through mobile apps [47,50,64].

A total of 6 Web-based interventions [25,33,34,38,44,59] were designed to provide on-demand information services and facilitate client-to-client communication. Web technologies are ubiquitous, flexible, and easy to deploy and have low operational costs [33,34,38,41,46]. They can offer a rich user-interaction through the intuitive presentation of information (eg, using multimedia or interactive content) [34,38]. On top of that, they can be tailor-made to fit the exact requirements of an intervention [33,34,38,50,60]. Intervention designers can use the Web as a supporting channel for parents to look up breastfeeding information, for example, a Web-based virtual maternity clinic [25], or as a Web-based discussion platform to connect mothers with care providers and peers [25,44,59].

In this group, 4 learning and training systems [23,41,42,44-46] and 4 client applications [35-37,39,40] were identified.

A total of 4 learning and training interventions [23,41,42,44-46] use Web platforms to improve breastfeeding knowledge of health care providers. In this context, Web-based systems become a potential learning platform because they can provide comprehensive sets of educational tools, such as lessons, knowledge evaluations [23,41], and Web-based discussion [44], and be adapted to the busy schedule of health care practitioners [41,46].

A total of 4 client applications [35-37,39,40] are self-service desktop programs to communicate breastfeeding information. These systems came in various shapes: interactive software running on touchscreen kiosks in clinics [39,68], software provided via CD-ROM [40], interactive agent that simulated lactation consultation [36], and breastfeeding quest game [37]. These experimental interventions aimed to explore the potential of interactive learning as an alternative mode of communicating breastfeeding information.

A total of 2 mobile apps [47,62,64] are categorized as geographic information systems, 2 Web interventions [25,60] are electronic medical record systems, and 1 mobile app [63,65] situates in both system categories.

FeedFinder [62] and MilkTrack [64], both of which also belong to the client communication system category, made use of positioning capability of smartphones to help mothers find and map suitable breastfeeding locations in public. In addition to breastfeeding location services, the MoomMae app [63,65] offers a breastfeeding diary function to keep track of mothers’ feeding habits, so they can accurately report feeding behaviors to health care professionals.

Both electronic medical record systems [25,60] are based on Web technologies. These interventions have an experience-sampling function that allows mothers to log their experience on a Web diary. The diary can be used as a personal record [25] or be accessed by health care providers to personalize their breastfeeding support tactics [60].

Telemedicine came in forms of remote lactation consultation via video calls [55-57] and email communication [58]. All 3 identified video-call interventions [55-57] took place in the United States, which has a geographically dispersed population. Video calls break distance barriers for parents who live in rural areas [55,57] and facilitate the delivery of professional breastfeeding support when health care facilities are limited [56]. More importantly, the technology allows rich communication in real time (eg, allows health care practitioners to observe breastfeeding parents and demonstrate the feeding process [55]). Email was also used by mothers to initiate communication with health care professionals and inquire for breastfeeding information [58].

This illustrates the breadth of information and communication systems that are available to support breastfeeding, both in terms of audience and underlying technology. Here, it is interesting to note that interventions predominantly rely on existing communication channels, such as SMS or phone calls, or implement straightforward and well-explored technologies, such as Web portals.

A total of 20 systems were evaluated in terms of usability, user satisfaction, and user experience. Usability and user experience of 12 systems were evaluated through user studies, whereas 8 clinical validations included measures of user satisfaction and discussions of usability flaws as a part of their evaluation process. Here, we organize the results of the studies according to their objectives.

Among the reported user studies, a breadth of quantitative and qualitative research methods was employed; participant samples included interaction design experts, health care experts, and prospective end users.

Usability was assessed with different methods, depending on the objectives of the evaluation, including expert reviews and user studies. The evaluation of MoomMae [63] involved interviews and surveys with 21 breastfeeding mothers who used the app for 4 weeks. Although the app was rated as useful, postexposure interviews revealed a negative usability trend; common flaws were confusing user interface elements, ease of use, and screen readability. The bilingual desktop breastfeeding education [35] was evaluated using Nielsen usability heuristics [69] by 2 usability experts, and the paper reported 91 usability flaws across 271 screens. Cheng et al [34] involved 20 participants in different places to evaluate a Web-based breastfeeding education system with graphical content in terms of user satisfaction and learning outcomes. The study shows that graphical content shortened view time, increased user satisfaction, and did not negatively affect learning.

Besides usability, factors contributing to user experience were also explored. The Milk Matters [47], Milk Man [61], and FeedFinder [62] apps, for example, involved end users to conduct formative tests throughout their design and development cycles to minimize usability issues and adapt to user preferences. Formative user studies with stakeholders were also conducted in the development of a culturally adapted Web-based intervention for indigenous mothers [33]. The evaluation of these systems stressed on other quality attributes, such as engagement, emotion, and ease of use in specific circumstances. The analysis of comments collected after an in-the-wild release of FeedFinder [62] app showed improvement of mothers’ confidence to breastfeed in public places. Engagement factors of the Milk Man app [61] included connectivity for fathers to seek and offer support, share personal experiences, and seek peer social connection. Likewise, the Growing Healthy Program [50] was found to be engaging because of high perceived usefulness, content that suits users’ parental beliefs, and— from a technical perspective—the use of push notifications.

Other quality measures concerned user satisfaction and perceived usefulness of the systems, although some of them did not report full-fledged user studies. In general, most of the interventions were perceived positively, but some unexpected usability flaws were discovered. For example, the CD-ROM desktop program [40] reported good reviews from participants; however, only 119 of the 240 participants in the intervention unit (a total of 993 mothers from both intervention and control groups combined) actually accessed the intervention. Likewise, participants in a 2-way SMS intervention [48] reported not being able to remember the predefined SMS response codes, suggesting that systems would need to be adapted for effective in-the-wild deployment.

A total of 16 systems (18 papers, see Multimedia Appendix 1) were validated in terms of clinical outcomes. Of those, we found 13 systems that fully or partially achieved their objectives, 1 system validation led to inconclusive results (not statistically significant), and 2 could not demonstrate a measurable effect. In Multimedia Appendix 1, we present a summary of the clinically validated systems and indicate reported effects on outcome parameters. Here, we organize systems based on their reported outcome measures.

The breastfeeding learning and training systems for health care professionals were effective with statistical significance in contributing to the knowledge of health care professionals and supporting skills, although selection bias might exist in the studies. The 3 in-the-wild evaluations [42,45,46] of the Breastfeeding Basics [43] educational program over the course of 9, 10, and 12 years with respective numbers of 15374, 18522, and 19671 participants using pre- and posttest consistently reported improvement in all aspects of breastfeeding knowledge among the included participants. It is worth mentioning that most of the participants included in these 3 studies were required to take part in the intervention for education or professional purpose. The Blackboard-based system [23] and the forum-based intervention [44] that were evaluated based on pre- and posttests with 86 nursing students and 42 pediatrics (recruited based on self-selection) were found to be effective in improving breastfeeding knowledge. Selection bias is also found in an evaluation of the Moodle-based intervention [41], with 15004 participants that showed positive effects on breastfeeding attitude and support practice; only participants who completed and passed the first-round evaluations were included in the study.

Some of the client communication systems were also effective and statistically significant in improving breastfeeding practice and determinants: an assessment of the clinic touchscreen kiosk breastfeeding intervention [39] with 46 mothers indicated an improvement in knowledge, self-efficacy, and intention to breastfeeding. A Web-based breastfeeding education [38] evaluated with 65 primigravids at their 29 to 36 weeks of pregnancy succeeded in improving breastfeeding rates, knowledge, and attitudes. A total of 2 [49,53] of the 5 SMS-based systems have reported statistically significant improvement of the duration of exclusive breastfeeding at the sixth month, with study cohorts of 582 and 298 participants. The evaluation of the postnatal phone-call support [51] with 178 new mothers through a follow-up phone call 3 months after the intervention commencement has reported a higher rate of exclusive breastfeeding and lower formula feeding rate among mothers in the intervention group, when compared with the control group. From a yearlong study with 414 mothers on multichannel lactation support [59], a positive long-term effect on exclusive breastfeeding in the intervention group was reported.

Not all effective systems fully achieved their goals. Despite showing significant improvements in breastfeeding duration and exclusivity, a study of the breastfeeding diary [60] with 96 participants did not show statistical significance in a second objective, decreasing postpartum depression. Similarly, a study with 24 participants on the Growing Healthy Program [50] has indicated an improvement in mothers’ confidence in their choice of feeding method (either breastfeeding or bottle feeding) and perception of milk supply. However, feeding plans and intentions were not influenced as the decisions were taken before exposure to the intervention. A study on an SMS-based breastfeeding encouragement system [48] with 234 participants was shown effective in improving breastfeeding exclusivity but showed no statistically significant impact on self-efficacy, which might offer an explanation for its failure to increase breastfeeding rates.

Finally, client applications did not produce measurable effects. A 1-month controlled trial with a total of 993 mothers (from intervention and control groups) on a CD-ROM software [40] and a 25-participant pilot study of the breastfeeding game [37] failed to improve knowledge [37,40], breastfeeding rates [40], intention [37], and self-efficacy [37]. A pilot controlled study of the interactive agent [36] with 15 women showed no significant improvement in breastfeeding intention and self-efficacy and failed to alter attitudes toward breastfeeding. It is worth noting that these complex systems were neither custom-built to suit the end users nor involved the primary stakeholders in the development process but rather a modification of existing technology platforms.

Here, the heterogeneous nature of interventions and validation processes and outcome parameters limit the opportunity to conduct a meaningful comparison between interventions. On a general level, our results suggest that breastfeeding learning and training systems for health care professionals tend to succeed in improving breastfeeding knowledge among their intended users, whereas some client communication systems showed a positive effect on improving breastfeeding adherence. Other breastfeeding determinants were rarely influenced, and validation studies evaluating complex technology (eg, games) suggested that such systems were ineffective in their current designs.