Impact of telelactation services on breastfeeding outcomes among Black and Latinx parents: protocol for the Tele-MILC randomized controlled trial

Breastfeeding offers numerous medical and neurodevelopmental advantages for birthing parents and infants, with increased benefits with longer breastfeeding duration [1‐4]. Breastfeeding includes feeding directly at the breast/chest as well as the provision of human milk (e.g., through pumping or expressing milk). Unfortunately, the majority of parents in the USA stop breastfeeding before it is recommended, with fewer than half of infants receiving any breastmilk at 6 months of age [5]. Breastfeeding rates are even lower among minoritized communities; for example, only 30% of Black infants receive breastmilk at 6 months [6]. Professional lactation support by the International Board Certified Lactation Consultants (IBCLCs) has been shown to improve breastfeeding duration and exclusivity (i.e., proportion of breastmilk in an infant’s diet) [7‐10]. As such, the Surgeon General’s Call to Action to Support Breastfeeding identifies increasing access to IBCLCs as a policy priority [11]. IBCLCs provide dedicated management of lactation problems that other providers (e.g., pediatricians) may not address because of limited training and competing demands [12, 13]. Poor access to IBCLCS among minoritized and rural parents may contribute to breastfeeding disparities [14].

Telelactation video visits may increase access to IBCLC services. IBCLC services are traditionally delivered in-person; however, telelactation, which allows remote IBCLCs to connect with breastfeeding parents via two-way video, is a potential alternative. Telelactation is less costly and more convenient than in-person office visits and allows new parents to avoid traveling with their infants. It can also address healthcare professional shortages, bringing IBCLCs into communities that lack them. In prior observational studies and case reports, telelactation was offered to augment in-person lactation support within the medical home or at IBCLC offices. Although informative and demonstrating promise, prior studies were small, used outdated technology, and assessed indicators of satisfaction and convenience rather than impacts on breastfeeding rates [15‐19]. Multiple companies [20] now offer a model of telelactation where patients can initiate video calls with lactation consultants using their personal devices (e.g., mobile phones) and access support on-demand at any time of day. The COVID-19 pandemic further accelerated the use of telelactation, with providers increasingly turning to virtual visits to maintain access to professional lactation support and avoid exposure to COVID-19. Despite the proliferation of these services, very little is known about the effect of telelactation. To date, no definitive evidence exists on whether these services improve breastfeeding rates and whether they are feasible and acceptable to different populations of parents.

In a pilot randomized trial of telelactation [21], our study team demonstrated the feasibility of on-demand telelactation and showed it was highly valued by new parents [22‐26]. Increased breastfeeding duration was observed among participants receiving telelactation, but the sample size was small and differences were not statistically significant. Also, the pilot occurred in a single, predominantly White community in rural Pennsylvania. A larger, more diverse study population is needed to quantify the impact of telelactation on breastfeeding outcomes, including the effects on subpopulations with historically low breastfeeding rates. The Tele-MILC (Telehealth for Mothers to Increase Lactation Confidence) trial described in this study protocol is designed to address this gap.

The trial has two aims and several sub-aims. The first aim is to assess the impact of telelactation on breastfeeding duration and exclusivity. Sub-aim 1A is to assess the effect of telelactation on breastfeeding outcomes across Latinx, Black, and non-Black and non-Latinx (predominantly White) parents and across rural and urban parents. Sub-aim 1B is to explore whether breastfeeding self-efficacy is a mediator for any observed improvements in breastfeeding outcomes. We hypothesize that a higher proportion of individuals with access to telelactation will be (1) breastfeeding at all and [2] breastfeeding with less formula use at 24 weeks postpartum and that these higher breastfeeding rates will be mediated by increases in breastfeeding self-efficacy. The second aim is to explore how acceptability of and experiences with telelactation vary across Latinx, Black, and non-Black and non-Latinx parents to guide improvements in these services.

The trial is a pragmatic, parallel-design randomized controlled trial. It is a digital trial, with all activities from recruitment to intervention delivery and outcome assessment conducted online. The overall study design is mixed methods and uses a sequential explanatory design in which qualitative interviews are used to explain and contextualize findings from the quantitative outcomes analysis.

In this fully digital trial, there are no physical study sites; however, study team members are located at RAND, University of Pittsburgh, and Harvard University. Recruitment occurs via Ovia’s pregnancy tracker mobile phone app. The Ovia app is one of the most popular pregnancy apps available for free download [27]. Used by over 2 million people in the USA each month, it provides educational content, conducts health assessments, and uses proprietary algorithms and machine learning to provide user-specific support, advice, and resources [27]. Study participants are drawn from a large community of active Ovia app users across the USA. Our focus will be on the subset of Ovia users residing in 39 US states and territories with fewer than 5 IBCLCs per 1000 births [28]. However, if recruitment occurs more slowly than planned, we will expand recruitment to Ovia users in all US states, and we will add Facebook as a recruitment site. With Facebook, we will contact the administrators of pregnancy and parenting groups and ask them to advertise the study to group members on our behalf.

Eligible parents include those who are (1) ≥ 18 years of age, (2) pregnant with their first child, (3) intending to attempt breastfeeding, and (4) residing in a state underserved by IBCLCs. Exclusion criteria include (1) non-singleton pregnancy, (2) advised by a healthcare provider not to breastfeed for a medical reason (e.g., HIV+ status, chemotherapy planned), (3) police custody or incarceration, and (4) infant to be separated from gestational parent (e.g., given up for adoption, surrogate, protective custody).

All potential trial participants provide informed consent in an eConsent process. The eConsent process features a short slide show which summarizes key points, a long form (detailed) consent statement, and an interactive quiz to probe understanding. The quiz is included to reinforce key concepts. Participants who enter incorrect answers are shown the correct answer with an explanation. Because incorrect answers receive this correction and reinforcement, incorrect answers do not prevent participants from continuing with consent and participating in the study. Participants sign the long form consent statement once they have reviewed all materials to their satisfaction and agree to participate.

There are no additional consent provisions because this is not applicable.

We plan to recruit 2400 participants over 27 months. To meet this target sample size, we first estimated the number of study participants needed to achieve 80% power in the primary, intent-to-treat (ITT) analysis, assuming a type 1 error rate of 5%. We applied national estimates of breastfeeding rates among diverse groups who had initiated breastfeeding and estimated that approximately 50–60% of control group participants would be breastfeeding at 24 weeks [31]. To detect a clinically significant difference of 10-percentage points in any breastfeeding at 24 weeks postpartum across the study arms (telelactation vs. usual care), we would need 355–390 participants in each study arm (710–780 participants total) if using either a test of equal proportions or testing significance of the “treatment” variable in a logistic regression model. This effect size corresponds to a clinically significant effect size within the range of published effect sizes for IBCLC interventions [10]. It is also within the range we expect to see given certain planned changes to the study protocol relative to the pilot study, which had between–group differences in rates of breastfeeding at 3 months post-partum of 3–11%.

Next, because we also plan to examine whether intervention effects vary across specific racial and ethnic groups (Black, Latinx, and non-Black and non-Latinx) and for rural vs. urban participants, we estimated the number of participants needed for this aim. For this analysis, we will fit a logistic regression model that includes the treatment dummy and a vector of race dummy variables, as well as their interactions. We assumed that breastfeeding rates at 24 weeks among control group participants are consistent with current CDC estimates (62% for non-Latinx and non-Black, 59% for Latinx, and 53% for Black participants [31]) and that we will secure equal numbers of participants in each combination of race/ethnicity and treatment. To achieve statistical power of 80% to detect heterogeneous treatment effects by race/ethnicity (e.g., effect size of 15% for Black, 15% for Latinx, and 0% for non-Black and non-Latinx participants; or effect size of 0% for Black, 0% for Latinx, and 15% for non-Black and non-Latinx participants), we would require 1800 participants. After inflating the sample size to account for 25% dropout, which is common for digital trials [32, 33], we would need to enroll approximately 2400 participants (800 in each subgroup). This sample size provides 80% power to detect an overall effect size as small as 6.5% across arms. If our final sample is representative of Ovia users in terms of rural vs. urban distribution, then 2400 participants will also provide sufficient power for testing whether treatment effects differ by rural vs. urban location. These sample sizes provide sufficient statistical power (>80%) for our additional planned analyses, including detecting the treatment effect in the instrumental variable analysis (which has similar sample size requirements) [34] and an additional survival analysis within Aim 1. Further, because the inclusion of covariates generally increases power, these estimates are conservative given the analysis plan.

Ovia sends emails featuring recruitment ads to subsets of the eligible population up to 3 times per month, and ads also appear in the within-app timeline of potentially eligible users. Ads we developed (Fig. 2), as well as all other study materials including draft surveys, were reviewed by a diverse advisory group of 18 parents who were recruited for this advisory group via Ovia. Materials were modified based on advisory group feedback prior to fielding.

Ads will run over 27 months and will be periodically suspended to pace recruitment (e.g., allow for participant verification). Throughout the recruitment period, we conduct ongoing monitoring for cases of fraudulent enrollment (e.g., bots rather than humans completing surveys, ineligible participants misrepresenting themselves, eligible participants attempting to enroll more than once) and investigate suspicious cases.

This protocol is Version 1 (09.12.2021). Recruitment began on July 7, 2021, and will be completed in October 2023.