Mental health disorders are among the most common complications during pregnancy and in the first 12 months after childbirth [1-3]. Research suggests that the prevalence of perinatal anxiety disorders is at least 17% and that approximately 7% to 20% of individuals experience clinical depression at some time during the perinatal period [4-6]. Perinatal mental health disorders make it difficult to function and care for oneself and for an infant. In fact, maternal mental health is considered an important underlying factor associated with barriers and reduced rates of human milk (HM) feeding intent, initiation, exclusivity, and continuation [1-3,7].

HM feeding is considered the ideal form of infant feeding because of its extensive benefits for the lactating person and infant. All major health and professional organizations recommend exclusive HM feeding for the first 6 months of a child’s life, with continued HM feeding in combination with appropriate complementary foods for at least 1 to 2 years [8-10]. However, despite the benefits, recommendations, and high rates of intention to HM-feed, overall rates within the United States continue to be low [11]. There are many known barriers to reaching HM feeding goals, including painful or difficult latch, concerns about milk supply, lack of professional lactation support, unsupportive social and cultural norms, inadequate parental leave policies, and maternal mental health difficulties [12-16].

Certain barriers are disproportionately experienced by Black individuals, many of which stem from historical and continued oppression, systematic racism, social injustices, and structural violence. For example, Black mothers often receive limited education and differential treatment from providers regarding HM feeding information and encouragement [12]. National data show that non-Hispanic Black mothers have the lowest rates of HM feeding initiation and continuation at 6 and 12 months postpartum compared with all other racial groups in the United States. [11] In addition, Black mothers are more likely to report depressed mood in late pregnancy and early postpartum than White mothers, even after adjusting for income and education, distinguishing between the effects of race and socioeconomic status [17].

The relationship between maternal mental health and HM feeding outcomes is bidirectional; mental health disorders can make HM feeding more challenging, and difficulty with HM feeding may predict depression and anxiety [18-22]. Therefore, it is important to consider both when designing interventions to improve these outcomes. Cognitive behavioral therapy (CBT), which focuses on identifying and changing unhelpful thoughts and behaviors, has been shown to be effective in preventing perinatal depression [23]. Interventions that extend across pregnancy and postpartum and offer individualized support from professionals and peers have been shown to be successful in improving both mental health and HM feeding outcomes [24,25]. In addition, intervention components shown to improve these outcomes among Black mothers include a positive representation that enhances and normalizes HM feeding in an encouraging way, content that addresses gaps in support (eg, building a support network, advocating for oneself in the hospital, preparing for a successful return to school or work, and enhancing HM feeding self-efficacy), and professional and timely HM feeding support that continues into the postpartum period [12,26-28]. Although many effective intervention strategies exist, access to these programs can be a barrier, especially for those navigating the complex and vulnerable early postpartum period.

The internet offers great potential in extending preventative and supportive services to individuals in the perinatal period as it addresses several key barriers to success. Digital technology interventions, which include the use of web-based content and interactions, SMS text messaging, and social media, have been effective in reducing depressive symptoms and improving HM feeding outcomes [29,30]. Black mothers report that social media, for example, is a practical, convenient, and valuable way of obtaining HM feeding information and support, feeling connected with people who have overlapping lived experiences, and improving self-efficacy [31,32]. However, there is limited evidence on the feasibility of such interventions for preventing perinatal mental health disorders and improving HM feeding outcomes in Black mothers.

The previously studied Sunnyside intervention is a web-based CBT-based internet program used to manage mood during the perinatal period. Findings from the pilot study showed that intervention use and satisfaction were high among participants, and symptoms of depression decreased from midpregnancy to 6 weeks postpartum [33]. To further study the relationship between mental health and HM feeding, we developed Sunnyside Plus, which is built upon Sunnyside and also includes HM feeding education and support. Therefore, the objectives of this study were 2-fold. First, we examined the feasibility of the Sunnyside Plus by measuring adherence to the intervention, usability, and acceptability. Second, we tested the preliminary mental health and HM feeding outcomes of Sunnyside Plus compared with those of Sunnyside.

This randomized pilot trial used a comparative effectiveness research approach to compare 2 active treatments, Sunnyside [33] and the newly developed Sunnyside Plus, on maternal mental health and HM feeding outcomes among Black individuals with mild to moderate depressive symptoms upon study enrollment who intended to HM-feed their child. Although HM feeding can include the use of donor HM, in this study, HM was provided directly from the lactating parent at their breast or chest or via their expressed milk.

Participants were recruited through advertisements placed on Ovia Health [34], a nationwide web-based pregnancy forum and internet-based application, between June 12, 2020, and September 15, 2020. Inclusion criteria were as follows: (1) pregnant and between 20 and 28 weeks of gestation, (2) aged ≥18 years, (3) Black or African American, (4) intending to HM-feed their child, (5) self-reporting mild to moderate depressive symptoms (Patient Health Questionnaire [PHQ]–8 score of 5-14), (6) access to a broadband internet connection, and (7) proficiency in the English language. Exclusion criteria were as follows: (1) pregnant with multiples; (2) visual, hearing, voice, or motor impairment that would prevent completion of the study procedures; (3) diagnosed with a major depressive episode, psychotic disorder, bipolar disorder, dissociative disorder, substance use disorder, or other diagnoses for which participation in this trial was either inappropriate or dangerous based on self-report; or (4) currently receiving treatment (medication or psychotherapy) and having an intention to resume antidepressant medication after birth (ie, those who discontinued their medication during pregnancy). Those interested were directed to a brief web-based screener to assess their eligibility.

Qualifying individuals provided electronic consent to participate. All procedures were approved by the institutional review board at the University of Illinois Chicago (UIC; IRB approval number: 2019-0519).

Following consent, qualifying individuals were immediately directed to complete the baseline assessment surveys that were provided via a REDCap (Research Electronic Data Capture; Vanderbilt University) link; study data were collected and managed using REDCap electronic data capture tools hosted by the UIC [35,36]. Participants were then randomized in a 2:1 allocation ratio to either Sunnyside Plus or Sunnyside using a block randomization method with a web-based randomization service provider, Sealed Envelope [37]. A 2:1 randomization allocation was used to gain more feasibility insights and experience with the Sunnyside Plus intervention component, which had not been previously studied.

All participants, regardless of group allocation, completed an initial engagement session to review the components and expectations of the study and ensure access to the treatment websites. The engagement session took place through the Cisco WebEx Meeting Center, a Health Insurance Portability and Accountability Act–compliant videoconferencing web application. Once completed, the web-based intervention began. Follow-up assessments using REDCap took place following the completion of 6 weeks of web-based lessons during pregnancy (third trimester) and at 6 and 12 weeks postpartum. A brief assessment of HM feeding continuation and exclusivity (yes or no reply) was performed on a weekly basis via SMS text messaging (SimpleTexting [38]) from 1 to 6 weeks postpartum. Participants received a US $20 Amazon gift certificate after completing each assessment.

Statistical analyses were performed using R software [46]. This study used a repeated measures design with 2 intervention groups (Sunnyside and Sunnyside Plus). Data were examined to assess for outliers. Descriptive statistics were obtained by computing means and SDs for continuous variables and frequencies for categorical variables. Significance testing for within- and between-group differences in mental health and HM feeding outcomes were assessed; however, this feasibility study was not powered for these types of analyses; therefore, our data are largely presented descriptively. These results should be interpreted with caution because of the small sample size. Differences in the baseline characteristics between the intervention groups were assessed using an independent 2-sample t test (continuous variables) and Fisher exact test (categorical variables). The intervention feasibility data, including adherence to the intervention, usability, and acceptability, were assessed using descriptive statistics.

In total, 1618 individuals (an average of 17 per day) completed the web-based screener across a 3-month period. Of these 1618 respondents, the mean age was 30.9 (SD 3.3) years, 539 (33.3%) were identified as Black or African American, 1171 (72.3%) intended to HM-feed, and the mean PHQ-8 was 6.2 (SD 4.7). Approximately 4.45% (72/1618) of individuals met the inclusion criteria. The major factors for exclusion were race and the estimated gestational age (EGA). Of the 1618 respondents, 874 (54.01%) of respondents identified as White. Only 38.44% (622/1618) of the respondents had an EGA between 20 and 28 weeks, which was the inclusion criterion for this study. Most had an EGA of <20 weeks (971/1618, 60.01%). Of those who qualified based on the eligibility screener, 31% (22/72) individuals were randomized and received the intervention; those who were not randomized were ultimately not interested in participating, unable to contact, or lost to follow-up. Further details are provided in Figure 1.

A total of 22 tertiary-educated Black pregnant individuals in their second trimester (mean EGA 22.6 [SD 2.5] weeks) participated in this US-based study. The mean age of the participants was 30.4 (SD 3.9) years. Most participants were married or partnered and cohabiting (17/22, 77%), employed full-time (13/22, 59%), or had private health insurance (16/22, 73%). All participants attended at least some college degree (12/22, 55%) or held a graduate or professional degree (10/22, 45%). Just over half (12/22, 55%) of the participants reported an annual household income of ≥US $51,000, with the average household size being 2.5 (SD 1). Maternal prepregnancy BMI (kg/m²) was calculated from self-reported height and prepregnancy body weight. When data from medical records were available, we explored the differences between self-reported and medical record data and found no differences. Over half of the participants (14/22, 64%) had obesity. The mean PHQ-9 score at baseline was 6.6 (SD 2.9), with most participants having mild depressive symptoms (15/22, 68%). The mean GAD-7 score was 6.05 (SD 4.7), with most patients having none to mild symptoms of anxiety (16/22, 73%). Most participants were nulliparous at enrollment (14/22, 64%). Regarding HM feeding–related variables, approximately one-quarter of participants reported being HM-fed themselves as an infant (6/22, 27%), most had no prior HM feeding experience (17/22, 77%), and most intended to HM-feed exclusively for at least the first few weeks postpartum (19/22, 86%). The PBSES score at baseline was 81 (SD 14.1), and the average HM feeding duration goal was 13.7 (SD 5.7) months. The participants in the intervention groups did not differ significantly in any baseline characteristics, suggesting that the random assignment–generated groups were equivalent at baseline (Table 2).

Approximately 78% (7/9) of Sunnyside participants and 77% (10/13) of Sunnyside Plus participants completed the study through 12 weeks postpartum. Approximately 9% (2/22) of participants (1 from each intervention group) withdrew from the study after the baseline assessment. One of the Sunnyside participants did not complete the third trimester follow-up assessment but continued with the study. One of the participants in the Sunnyside Plus group did not engage in the web-based intervention or complete the assessments and was therefore withdrawn by the research team after the baseline assessment. Approximately 9% (2/22) of additional participants (1 from each intervention group) were considered lost to follow-up during the postpartum period; their mood scores did not differ from those who completed the study.

Adherence to the web-based intervention was measured by the number of log-ins to the site during the intervention period, number of lessons accessed, and number of tools completed. Table 3 shows site use data. The mean number of log-ins across the 6-week intervention plus booster sessions was 7.3 (SD 5.3) for Sunnyside and 13.8 (SD 10.5) for Sunnyside Plus. Within the Sunnyside group, the average number of lessons accessed during pregnancy (from a total of 13) was 10.1 (SD 3.5) and during postpartum (from a total of 3) was 1.6 (SD 1.3). Within the Sunnyside Plus group, the average number of lessons accessed during pregnancy (from a total of 13) was 9.7 (SD 4.1) and during postpartum (from a total of 9) was 2.6 (SD 3.3). Approximately 67% (6/9) of Sunnyside participants and 58% (7/12) of Sunnyside Plus participants completed at least 50% of the available lessons. The average number of tools used was 11 (SD 6.6) for Sunnyside and 25.8 (SD 27.8) for Sunnyside Plus. Participants in the Sunnyside Plus group used the activity tool more than those in the Sunnyside group (P=.03). All other site uses were similar between the groups, with no additional differences found.

Mean number of weekly HM feeding text question responses across the first 6 weeks postpartum was 5.4 (SD 1.1) for Sunnyside and 5.2 (SD 1.3) for Sunnyside Plus. Approximately 56% (5/22) of participants in Sunnyside and 46% (6/13) of participants in Sunnyside Plus completed all 6 weekly text questions. Response rates were similar between the groups, and no differences were found. Participants in the Sunnyside Plus group were offered web-based lactation support. The mean number of video calls completed during the first 6 weeks postpartum was 2.8 (SD 2). The number of calls ranged from 0 to 7. One of the participants declined to receive lactation support. The results are presented in Table 3.

At the third trimester follow-up (after completion of the 6-week antenatal web-based intervention), scores on USE subscales ranged from 1 (strongly disagree) to 7 (strongly agree). Mean scores for the participants in the Sunnyside group were 5.3 (SD 1.3) for usefulness, 5.1 (SD 2.1) for ease of use, 5.1 (SD 2.3) for ease of learning, and 4.9 (SD 1.9) for satisfaction. Mean scores for the participants in the Sunnyside Plus group were 4.9 (SD 1.2) for usefulness, 5.9 (SD 1.1) for ease of use, 6.2 (SD 1.1) for ease of learning, and 5.3 (SD 1.5) for satisfaction.

At 6 weeks postpartum, the scores on the USE subscales also ranged from 1 (strongly disagree) to 7 (strongly agree). Mean scores for the participants in the Sunnyside group were 4.9 (SD 1.4) for usefulness, 5.4 (SD 1.3) for ease of use, 5.5 (SD 1.5) for ease of learning, and 4.7 (SD 0.7) for satisfaction. For participants in the Sunnyside Plus group, mean scores were 5.1 (SD 1.4) for usefulness, 6 (SD 1.3) for ease of use, 5.8 (SD 1.3) for ease of learning, and 5.1 (SD 1.5) for satisfaction. The data are provided in Table 4. Overall, the usability and acceptability scores in this trial were slightly higher than those in previously published pilot data on Sunnyside [33].

Total gestational weeks at birth, provider type, and mode of birth did not significantly differ between the groups. Mean EGA at birth was 38.7 (SD 1.3) and 38.3 (SD 1.7) for Sunnyside and Sunnyside Plus participants, respectively. A reported 71% (5/7) and 73% (8/11) of participants in Sunnyside and Sunnyside Plus, respectively, received care under an obstetrician rather than a midwife. Approximately 43% (3/7) of Sunnyside and 64% (7/11) of Sunnyside Plus participants had a cesarean birth.

Mental health outcomes are presented in Table 5. No differences between or within the groups were detected on any of the mental health outcome measures. In both intervention groups, mean PHQ-9 scores across all follow-up visits remained <10, which is the clinical threshold for referral for treatment [40]. Mean IDAS scores remained relatively consistent across all follow-up visits for Sunnyside and Sunnyside Plus. In both intervention groups, mean GAD-7 scores across all follow-up visits remained <10, which is the threshold for moderate to severe symptoms of anxiety [42].

Prenatal HM feeding outcomes were examined using descriptive statistics (data not shown). The mean intended HM feeding duration at baseline (midpregnancy) was 12.7 (SD 6.2) months for Sunnyside participants and 14.3 (SD 5.5) months for Sunnyside Plus participants. Among all participants, 86% (19/22) intended to HM-feed exclusively for at least 5 to 6 months. Baseline prenatal HM feeding self-efficacy (PBSES) scores were 77.3 (SD 14.3) for Sunnyside participants and 83.5 (SD 14) for Sunnyside Plus participants. The PBSES scores increased slightly for all participants from baseline to the third trimester.

Postpartum HM feeding outcomes are shown in Table 6. All participants initiated HM feeding treatment (18/18, 100%). At 6 weeks postpartum, 57% (4/7) of Sunnyside and 91% (10/11) of Sunnyside Plus participants reported HM feeding exclusively. At 12 weeks postpartum, 57% (4/7) of Sunnyside and 80% (8/10) Sunnyside Plus participants were exclusively HM feeding. Any HM feeding at 6 weeks postpartum was reported by 86% (6/7) of Sunnyside and 100% (11/11) of Sunnyside Plus participants. Similarly, any HM feeding at 12 weeks postpartum was reported by 86% (6/7) of Sunnyside and 100% (10/10) of Sunnyside Plus participants. No differences were detected in the postpartum HM feeding outcome measures between the groups.

This study describes the feasibility and preliminary findings of a novel CBT-based internet intervention to prevent perinatal depression and promote HM feeding in Black mothers with mild to moderate depressive symptoms and intention to HM-feed. Although both active treatment groups aimed to target skills to manage mood, the newly developed Sunnyside Plus intervention used evidence-based practices to promote and actively support the HM feeding.

Participants for this study were enrolled within a relatively short recruitment period (3 months) of indicating interest in the study intervention. Relatively low attrition rates in the Sunnyside group (2/9, 22%) and in the Sunnyside Plus group (3/13, 23%) through 12 weeks postpartum in a population of individuals within the perinatal period and who indicated mild to moderate depressive symptoms suggested adherence to the intervention. Further adherence was shown through participant interactions with the site through log-ins, lessons accessed, and tool use and with the study team through SMS text message responses and lactation support calls. Adherence to the web-based intervention indicates that Black individuals who are in the perinatal period are willing to use an individual intervention program that involves engagement with a website and interactions via SMS text messaging and video calls. To improve adherence, future research efforts using the Sunnyside intervention should consider adding biweekly support calls with participants to remind them of the intervention components and resolve any potential access issues. Usability scores suggested an overall positive user experience for both the pregnancy and postpartum sections of the intervention.

After completion of the intervention at both 6 and 12 weeks postpartum, no participants in this at-risk sample met the criteria for postpartum depression. This is in line with results from the previously published pilot data on Sunnyside [33]. Other studies have shown a 13% prevalence rate of postpartum depression among individuals in the first year postpartum [6] and a 17% prevalence rate among at-risk individuals in the absence of an intervention [47]. Overall, the levels of perinatal anxiety symptoms remained low among the participants. Given the adverse impact of perinatal mental health disorders on both the mother and infant, including reduced rates of HM feeding [3], the overall low levels of depression and anxiety symptoms among all participants were encouraging.

In the United States, 80% of mothers intend to HM-feed in some capacity, and of those, >85% intend to exclusively HM-feed for at least 3 months; however, only one-third (32%) of mothers achieve their intended HM feeding goals [48]. In this study of individuals who intended to HM-feed in at least some capacity, intended HM feeding exclusivity and duration were high among all participants at baseline. Postpartum HM feeding self-efficacy, defined as the confidence in one’s ability to effectively HM-feed, is thought to play an important role in the relationship between postpartum depression and HM feeding [19,49-52]. Not only is high self-efficacy associated with lower levels of depressive symptoms [45,50,51] but also with longer HM feeding durations [50,51]. In this study, prenatal HM feeding self-efficacy was high among all participants, and scores increased after completion of the antenatal portion of the intervention.

According to the National Vital Statistics System, the US cesarean birth rate in 2019 was 32% [53]. Overall, 56% (10/18) of the participants in this study had a cesarean birth, which is almost double the national rate. Medical interventions during birth, including cesarean birth, may make it difficult for mothers to reach their HM feeding goals [15]. Across both intervention groups, 100% (22/22) of participants initiated HM feeding. The HM feeding initiation rate among Black individuals in the United States is 60% [54]. At 6 weeks postpartum, a greater percentage of Sunnyside Plus participants exclusively HM-fed than those in Sunnyside (10/11, 91%, vs 4/7, 57%, respectively). At 12 weeks postpartum, a greater percentage of Sunnyside Plus participants exclusively HM-fed than those in Sunnyside (8/10, 80%, vs 4/7, 57%, respectively). Most participants reported at least some HM feeding at both 6 and 12 weeks postpartum (6/7, 86%; 11/11, 100%, or 10/10, 100%, for Sunnyside and Sunnyside Plus, respectively). These rates are higher than that of research showing an 81% prevalence rate of any HM feeding at 6 weeks postpartum [51]. High rates of exclusive and continued HM feeding at 6 and 12 weeks further underscore the positive clinical impact of both intervention groups.

There are several strengths of this study. The design of the intervention offers a novel approach for preventative and supportive care within the perinatal period, one that extends across pregnancy and postpartum, involves various interface options (ie, website, text, and videoconferencing), and acknowledges the logistical challenges of physically seeking care as parents of a newborn. In addition, participants consistently used all aspects of the intervention in both pregnancy and the postpartum period, suggesting interest in and satisfaction with this design of care. To reduce the race-mediated power differential, the lactation specialist team included Black and White individuals. When support was provided by a White lactation specialist, we acknowledge that race-of-interviewer effects may have been present.

This study had several limitations. Although the recruitment response was high, with an average of approximately 17 respondents each day during the 3-month recruitment period, the use of a convenient internet sample may have led to a bias toward higher education levels. In addition, our recruitment method through Ovia Health might not have been the best route, given the characteristics of those using the platform, and future projects should consider other recruitment routes. The major factors for exclusion were race and EGA. Future recruitment efforts should target internet-based applications used by those who identify as Black or African American. In addition, a system that allows for future rescreening might capture those who meet all inclusion criteria, except for the current EGA. As a small pilot study, this trial was not powered to reliably detect small significant differences or associations. The results should be interpreted with caution, and a larger trial is needed to verify these outcomes. Furthermore, the primary outcome data were based on self-report assessments, which may introduce recall bias. Finally, with no true control group, we relied on outside data to compare the rates of mental health and HM feeding outcomes.

The results of this study suggest that tertiary-educated Black mothers at risk for perinatal depression and who intended to HM-feed were receptive to, engaged with, and satisfied with a web-based CBT-based internet intervention, with and without HM feeding education and support, spanning from midpregnancy through 6 weeks postpartum. Preliminary findings indicate that both Sunnyside and Sunnyside Plus interventions have the potential to affect symptoms of depression, anxiety, and HM feeding outcomes. Future studies should include a larger sample size and a longer follow-up period to better understand the differences between groups and examine the continued impact across the postpartum period.