The use of digital communication platforms to improve health behaviors has expanded dramatically in the last decade. Digital media and marketing technologies such as SMS or text messages, mobile apps, and social media are being leveraged to reach diverse populations in ways that are efficient, convenient, and engaging [1-4]. These technologies are being used to promote healthy behaviors by answering client questions in real time, monitoring health behaviors, and providing education and support [4-7]. Digital interventions are often low cost and relatively easy for public health organizations to implement compared with in-person interventions [8]. They have frequently been used to target specific populations, including young people (because they are the most frequent users of smartphones and social media platforms) and low-income individuals who may not have consistent access to high-quality health information and may struggle with access to in-person preventive health services owing to transportation challenges, lack of childcare, and other barriers [1,9-13].

Digital communication campaigns have been leveraged to improve nutrition and physical activity (PA), increase breastfeeding and immunization rates, reduce barriers to preventive health screenings, and improve access to mental health support [14-17]. Petkovic et al [14] reported that social media interventions designed to increase PA may be effective, and social media interventions may improve overall well-being. However, the review indicated that there was insufficient evidence for improvement in other outcomes, including breastfeeding, condom use, diet quality, medication adherence, medical screening and testing, tobacco use, and vaccination. Welch et al [15] provided an overview of systematic reviews of social media intervention studies targeting a broad range of health behaviors and reported mixed effects on health behaviors and outcomes. The authors reported that social media interventions are more effective among disadvantaged populations such as youth, older adults, low-income people, and people residing in rural areas. Richardson et al [16] reported that SMS text messaging is an effective tool to influence caregiver behavior and child outcomes in health, whereas the results of the studies reviewed by Poorman et al [17] were mixed, and the authors concluded that SMS texting interventions that are based on an established theory of behavior change and use motivational language, as opposed to informational language, are more likely to be successful.

To date, no reviews have examined the use of different digital communication platforms to improve health behaviors among vulnerable families. Accordingly, this review focused on the use of diverse digital health promotion campaigns by community-serving agencies to reach low-income pregnant women, parents of children aged <18 years, and adolescents living in high-income, predominantly English-speaking countries. Members of the selected population share demographic, income, and health characteristics in addition to their use of community support services. For the purposes of this study, community-serving agencies included health departments; local nutrition agencies such as the Supplemental Nutrition Assistance Program and Special Supplemental Nutrition Program for Women, Infants, and Children (WIC); food pantries; public schools; home-visiting programs; and early childhood education programs. Considering the extensive use of digital technology among young parents and adolescents; the key developmental periods associated with pregnancy, childhood, and adolescence; and the barriers associated with accessing preventive health services in person, digital communication platforms are particularly relevant for this population [17-19].

Although existing reviews have focused on the correlation between the use of these technologies and health behavior change, this review also incorporated an investigation of the degree of engagement with and acceptability of these platforms [20]. It is important to explore technologies that elicit the greatest engagement from members of this population, as well as those that are the most acceptable, because these are important factors in the relationship between digital technologies and behavior change and could provide insights into the inconsistent impacts of digital health promotion campaigns on health behaviors [20-24].

The objective of this review was to identify the types of digital communication strategies used by community-serving organizations to promote healthy behaviors among vulnerable families, assess the strength of evidence for health behavior change, and describe the degree of engagement with and acceptability of various strategies.

The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines were followed in this review [25]. The PRISMA checklist is available in Multimedia Appendix 1.

Literature searches were conducted using 5 electronic publication databases (PubMed, Scopus, Web of Science, CINAHL, and APA PsycInfo). In addition, the lists of references of relevant systematic reviews and reports were searched for studies relevant to the review.

This review used a systematic approach to retrieve the relevant research studies and included all study designs for identifying the wide range of evidence examining the use of digital communication (ie, texting, mobile apps, or social media) to promote health behavior change among the target population. The inclusion criteria are detailed in Textbox 1. For the purposes of this review, program participation and use of program components were considered health behaviors. Studies that discussed industry marketing were excluded.

The participants were low-income parents and caregivers of children or adolescents, pregnant women, and adolescents participating in community-serving programs.

The initial search was performed between June 2020 and September 2020. A subsequent search was conducted in April 2022 to ensure the inclusion of recent publications. Search terms relevant to 3 distinct concepts were used, as outlined in Textbox 2. All titles and abstracts were initially screened to ensure relevance, no duplication of articles, and adherence to the inclusion criteria.

The retrieved studies were imported into the screening and data extraction tool developed by Covidence, an Australian nonprofit [26], for title and abstract screening, full-text review, and extraction. Of the 335 nonduplicated records, 62 (18.5%) articles were subjected to full-text review by 2 authors to determine eligibility (Figure 1). The reasons for excluding studies at the full-text review stage included the following: the study did not use digital communication to improve health behaviors; the study was not led by a community-serving agency; the study did not target the population of interest; and the results of the study were published elsewhere. A total of 23 articles were eligible for data abstraction and were included in this review.

The purpose of the digital campaign was first identified using an emergent process. The broad categories of purposes included health promotion and education, support, and behavior monitoring; more specific categories were determined during the analysis and are described in the Results section.

The degree of participant engagement and acceptability associated with each digital health promotion campaign was then assessed. In digital media studies, some forms of engagement can be measured across different platforms, whereas others are specific to a particular platform. For example, enrollment can be assessed for texting, mobile apps, and social media; responding can be assessed for texting and apps; clicking can be assessed for texting (of embedded links) and social media; and frequency of use can be assessed for apps and social media. Platform-specific assessments include reading (texting only), comments (social media only), likes (social media only), shares (social media only), and the use of app components (mobile apps only). To account for the differences in measurements among the platforms, the authors first categorized the studies by platform and then conducted an analysis of engagement. Across the platforms studied, acceptability measures included asking users about digital campaign usefulness, convenience, understandability, and enjoyability.

Finally, the quality of the health behavior evidence described in the studies was appraised by 4 researchers. The risk of bias was determined using the Effective Public Health Practice Project Risk of Bias tool [27] and included measurements for selection bias, confounders, reliable data collection tools and procedures, sample retention (ie, attrition rates), and protocol integrity. The strength of evidence assessments were conducted using criteria from the Agency for Healthcare Research and Quality [28], which included study limitations, directness of intervention, consistency and precision of implementation, reporting bias, and risk of bias. Each study was assessed independently by 2 researchers, first for risk of bias and then for strength of evidence. Any disagreements in assessments were reconciled by a third researcher.

The sample of this review consisted of 23 academic studies published between 2012 and 2022. Most (21/23, 91%) studies were conducted in the United States; 1 study was conducted in the United Kingdom, and 1 in Australia.

Of the 3 digital media platforms under review, the sample included 12 (52%), 6 (26%), 3 (13%), and 2 (9%) studies exploring the use of texting, mobile apps, social media, and multiple platforms (eg, texting and mobile apps), respectively. Of the 12 texting studies, 5 (42%) explored the use of one-way texting (ie, no opportunity for participants to respond), whereas 7 (58%) used a two-way texting campaign (participants could respond to investigators). Of the 23 studies, 1 (4%) study used both social media and an app, and 1 (4%) used both texting and an app. The social media studies focused on Facebook (3/23, 13%) and Twitter (1/23, 4%). The targeted health behaviors included healthy diet, PA, healthy weight and obesity prevention (9/23, 39%), healthy pregnancy (5/23, 22%), breastfeeding (4/23, 17%), vaccine use and disease prevention (2/23, 9%), smoking cessation (1/23, 4%), and WIC benefit redemption (1/23, 4%). Overall, 4% (1/23) of studies did not focus on health behavior but explored the feasibility of using social media to communicate with low-income mothers participating in a home-visiting program [29]. Community-serving agencies included WIC (12/23, 52%), health departments (2/23, 9%), food pantries (1/23, 4%), K-12 schools (2/23, 9%), public housing (1/23, 4%), home visits (1/23, 4%), Healthy Start (1/23, 4%), and a combination thereof (3/23, 13%).

The 23 study designs included 8 (35%) randomized controlled trials (of which 3, 37% were randomized controlled pilot studies), 6 (26%) pretest-posttest design, 3 (13%) mixed methods studies, 2 (9%) pilot studies, 1 (4%) feasibility study, 1 (4%) single-sample prospective cohort study, 1 (4%) descriptive study, and 1 (4%) cross-sectional study. The sample sizes ranged from 14 [30] to 30,440 [31] with a median of 77.5. The study population consisted primarily of reproductive-aged women (12/23, 52%), including those who were pregnant or mothers, caregivers of infants and young children (7/23, 30%), and adolescents (2/23, 9%). One of the 23 studies (4%) focused on mother and child dyads, wherein the child was 9-14 years old. [32]. The following data for each study are presented in Table 1: digital media platform, study reference, targeted health behavior, study objective, sample size, community agency, location, study design, theoretical underpinnings, description of the intervention, and duration of the study.

Each digital campaign could have one or more purposes, including general health messages, individually tailored messages, reminders, client support, or behavior monitoring. Individually tailored messages were sent to study participants based on certain characteristics, such as trimester of pregnancy, whereas general health messages were sent to all study participants regardless of their individual characteristics. General health message topics included healthy pregnancy, breastfeeding, healthy diet, fruit and vegetable (FV) consumption, PA, obesity prevention, vaccine use, and disease prevention, whereas individually tailored message topics included healthy pregnancy, smoking cessation, breastfeeding, and healthy weight. Campaigns whose purpose included sending reminders prompted study participants to engage in certain health behaviors at specific times. The remaining topics included health care appointments, healthy diet, and PA. Campaigns centering on support used the digital platforms to assist study participants with a particular health goal, such as smoking cessation, healthy diet and PA, breastfeeding, healthy pregnancy, disease prevention, or engagement in a home-visiting program. Finally, behavior monitoring involved the use of the digital platform to track health behaviors and outcomes, including PA and body weight.

The purpose of the digital campaigns included general health messages only (8/23, 35%), individually tailored messages only (2/23, 9%), general health messages and reminders (1/23, 4%), general health messages and support (3/23, 13%), general health messages and behavior monitoring (2/23, 9%), individually tailored messages and support (2/23, 9%), individually tailored messages and behavior monitoring (1/23, 4%), general health messages, reminders, and support (1/23, 4%), individually tailored messages, reminders, and support (1/23, 4%), and general health messages, reminders, support, and behavior monitoring (2/23, 9%; Multimedia Appendix 2). Of the 23 campaigns, 17 (74%) campaigns used general health messages, 6 (26%) used individually tailored messages, 5 (22%) sent reminders, 9 (39%) provided support, and 5 (22%) conducted behavior monitoring. Two-way texting campaigns were much more likely to be designed to provide support (6/23, 26%) than one-way texting campaigns (1/23, 4%). Behavior monitoring was conducted using mobile apps (3/23, 13%) and two-way texting (2/23, 9%).

Engagement with and acceptability of the platforms, as reported in 87% (20/23) of the studies, were concrete indicators of study participant behavior and are important mediators of health behavior change. Engagement and acceptability results are reported in the following sections, followed by health behavior change outcomes. Multimedia Appendix 3 includes a summary of engagement, acceptability, and health behavior outcomes.

Meta-analysis regarding health behavior outcomes was not completed because of the high degree of heterogeneity among the studies. Of the 18 studies that measured the impact on one or more health behaviors, 15 (83%) studies achieved at least 1 of their health behavior objectives, and 6 (33%) achieved all of their objectives. However, owing to the high prevalence of pilot studies, pretest-posttest design study designs, and relatively small sample sizes (median 77.5), the strength of evidence related to health behavior change was generally low and the risk of bias was moderate to high for the studies in this review (Multimedia Appendices 4 and 5). Additional factors leading to high risk of bias included a lack of valid and reliable data collection tools and high attrition rates. The strength of evidence was, on average, insufficient in the social media studies, low in the texting studies, and low to moderate in the mobile app studies and the studies exploring multiple platforms.

This study presents a review of digital communication campaigns used by community-serving agencies to promote health behavior change among vulnerable families. Furthermore, the study reviews their purpose, platform, and key methodologies and systematically describes the engagement with and acceptability of the respective campaigns among this population. Most of the community-serving organizations were WIC agencies, followed by health departments and K-12 schools. The platforms of interest in these studies included SMS text messaging, mobile apps, and social media; and most campaigns promoted healthy eating and PA, breastfeeding, and healthy pregnancies. The results of this review suggest that there is significantly more research focusing on the use of SMS text messaging to promote healthy behaviors in this population than the other 2 platforms and is consistent with prior studies [23]. This could be explained by the relative time and cost efficiency of implementing a texting campaign, the widespread use of texting (compared with the use of social media and mobile apps), and the alignment of texting with existing practices at community-serving agencies (ie, communicating by phone with clients).

The findings of this review related to engagement with and acceptability of these platforms are important considerations for community-serving agencies exploring the use of digital media campaigns to improve their clients’ health. All 3 platforms (9 texting studies, 5 app studies, and 2 social media studies) were highly acceptable, including when combined, with no notable differences in acceptability between the platforms. Study participants appreciated the convenience, short length of the messages (for texting and app studies), usefulness of the information received, and reminders (in texting studies). In addition, the capacity for interactive dialogue when available was particularly appreciated, as was the ability to create content.

Compared with acceptability, the engagement results were less consistent. Engagement was moderate with texting (7/23, 30% studies) and mobile app campaigns (6/23, 26% studies) and low with social media campaigns (3/23, 13% studies). In social media studies, engagement was higher at the beginning of the study period but tapered off over time. Campaigns with tailored messages and opportunities for interaction between study participants and providers tended to elicit greater engagement. This finding is supported by prior research that demonstrates that highly tailored interactive messages (eg, two-way texting and providing opportunities for client feedback) tend to be more successful [13,52,53]. In this review, 9% (2/23) of the studies explored the use of two-way texting for peer support, specifically support from breastfeeding peer counselors [42,43]. Similarly, Dion [29] described the benefits of peer support on a Facebook page used for home visiting services for new mothers. These studies provide further evidence for the high acceptability and appreciation for digital support from peers, which has been demonstrated by Fortuna et al [54] in a 2020 review of digital peer support provided by mental health professionals. Fortuna et al noted that digital support enables asynchronous communication and can facilitate wider reach for peer health professionals.

There was limited evidence to suggest that these digital media campaigns were effective in improving all targeted health behaviors; however, most interventions were successful or partially successful in improving some targeted health behaviors. The strength of evidence was generally low because of the high number of pilot studies, pretest-posttest study designs, and relatively small sample sizes in the study sample. The evidence base could be strengthened by a greater number of randomized controlled trials, as well as studies with larger sample sizes, such as prospective cohort studies. Triantafyllidis et al [55] similarly called for more high-quality studies of mobile health interventions. Considering the COVID-19 pandemic and increased use of digital communication across public health programs, high-quality studies of these platforms are urgently needed.

A key consideration for community organizations implementing digital health campaigns is the need to protect the data security of clients. Web-based consumer data are generally not well protected [7]. The use of mobile apps, self-tracking devices, and social media sites can render people vulnerable to web-based predators. This consideration was not adequately addressed in the studies in this review. Community organizations should take measures to protect their clients’ data privacy, and more research is needed to identify best practices for safeguarding the personal data of health care and social service agency clients.

Another important consideration is the disparities that exist in access to and use of digital communication technology and varying levels of digital health literacy among low-income groups, immigrants to the United States, and people with lower educational attainment (known as the “digital divide”) [17,56-58]. The studies in this sample generally avoided issues related to access by excluding participants without a mobile phone or, in the case of 6 studies, providing a mobile phone with a cellular or data plan to study participants. This could jeopardize the generalizability of the findings related to engagement and acceptability because the sample of this review is likely more digitally connected than the broader population of pregnant women, parents, caregivers, and adolescents.

With regard to digital health literacy, research suggests that one way to help ensure that digital communication campaigns are accessible to people with lower levels of digital health literacy is to actively involve members of the target population in the design of messages and other forms of content [4]. In this review, 9 studies described using participatory designs, including pilot testing, interviews, focus groups, surveys, usability testing, and convening diverse advisory boards. These designs are promising and should be integrated into a greater number of digital health communication studies.

This review has several additional limitations. First, the body of studies was heterogeneous with regard to the study design, study objectives, sample size, and measures of health outcomes. Furthermore, there were no clear definitions of platform engagement in the included studies and very little consistency in how engagement was measured and reported, making it difficult to compare studies even within 1 digital platform. Although the authors were able to summarize the state of the literature, this heterogeneity prevented meta-analysis and made it difficult to generalize the findings. Second, the lack of standardized keywords for the third concept in the literature search may have led to some studies that matched the criteria being excluded inadvertently. Terms such as “mHealth,” “text message interventions,” “social media–based interventions,” “digital technology,” “digital health,” and “digital health and technology intervention” were not included in the search. Future reviews of this subject matter should consider a broader set of search terms. Third, although the authors originally intended to include studies from a gray literature database, the study that met the inclusion criteria was ultimately considered anomalous from the peer-reviewed studies and excluded from the sample. The exclusion of gray literature may have precluded the authors from exploring certain relevant themes, as many of the studies that are conducted by or involve community organizations are less formal and are not reported in peer-reviewed journals.

This study was the first systematic review of the use of different digital media platforms by community-serving agencies. Digital media platforms are an important component of the health promotion activities of community-serving agencies in high-income English-speaking countries. Although the effectiveness of these strategies in improving health behaviors was inconclusive, low-income pregnant women, parents and caregivers of young children, and adolescents demonstrated moderate levels of engagement with and high levels of acceptability of these strategies.

Higher-quality studies with larger sample sizes are needed to strengthen the evidence base for the use of these platforms to improve health behaviors. Likewise, more research is needed to consistently measure and report participant engagement within and across platforms to discern which platforms and strategies are best suited for specific populations and purposes. This information could be helpful for community-serving agencies in their decision-making regarding which platforms and strategies would best suit their clients’ needs. Finally, future studies should consider strategies for protecting clients’ personal data, and more studies should involve members of the target population in the design of messages and content.

This review was not registered, and a review protocol was not prepared.