Breastfeeding and the origins of health: Interdisciplinary perspectives and priorities

Abstract Breastfeeding and human milk (HM) are critically important to maternal, infant and population health. This paper summarizes the proceedings of a workshop that convened a multidisciplinary panel of researchers to identify key priorities and anticipated breakthroughs in breastfeeding and HM research, discuss perceived barriers and challenges to achieving these breakthroughs and propose a constructive action plan to maximize the impact of future research in this field. Priority research areas identified were as follows: (1) addressing low breastfeeding rates and inequities using mixed methods, community partnerships and implementation science approaches; (2) improving awareness of evidence‐based benefits, challenges and complexities of breastfeeding and HM among health practitioners and the public; (3) identifying differential impacts of alternative modes of HM feeding including expressed/pumped milk, donor milk and shared milk; and (4) developing a mechanistic understanding of the health effects of breastfeeding and the contributors to HM composition and variability. Key barriers and challenges included (1) overcoming methodological limitations of epidemiological breastfeeding research and mechanistic HM research; (2) counteracting ‘breastfeeding denialism’ arising from negative personal breastfeeding experiences; (3) distinguishing and aligning research and advocacy efforts; and (4) managing real and perceived conflicts of interest. To advance research on breastfeeding and HM and maximize the reach and impact of this research, larger investments are needed, interdisciplinary collaboration is essential, and the scientific community must engage families and other stakeholders in research planning and knowledge translation.


| WORKSHOP RATIONALE AND METHODS
Breastfeeding provides a constellation of health benefits for mothers and infants. Considering the abundance of evidence and long-standing global recommendations to support breastfeeding, it is surprising that we still do not understand the underlying biological mechanisms of these benefits, and it is concerning that most mother-infant dyads do not achieve breastfeeding recommendations. To address these issues, we convened a workshop of experts in the field of breastfeeding and human milk (HM). The workshop focused on two main areas of concern emphasized by participants through priority-setting exercises before and during the workshop: the need for more interdisciplinary research in this field and the need to address counterproductive tensions between breastfeeding research and advocacy efforts.

| Counterproductive tensions
Like other areas of study, the major sources of funding for research on HM and infant feeding originate from governments, philanthropic or charitable foundations and other non-profit organizations, and industry. Many HM and breastfeeding researchers carefully manage potential conflicts of interest (COIs) with industry. Others choose to avoid financial COI altogether, and some also recognize and uphold the World Health Organization (WHO) Code of Marketing of Breastmilk Substitutes, which, although relevant to breastfeeding, is focused on the marketing of commercial products and not research governance. Scientists on both ends of this spectrum have been publicly shamed for their decisions. Aside from directly impacting the targeted individuals and areas of investigation, these dynamics may discourage young scientists from entering the field. Breastfeeding and HM researchers must also navigate increasingly complex social challenges when translating their research because social media is increasingly used to perpetuate misinformation or biased interpretations of scientific evidence about breastfeeding and infant formula. Correcting misinformation is challenging and time consuming and can detract from research activities. Although these challenges are not entirely unique to breastfeeding and HM research, they are heightened in this field due to the emotion associated with infant feeding decisions.

| Methods
To discuss these challenges facing breastfeeding and HM researchers, a multinational group of breastfeeding and HM researchers from diverse disciplines and career stages gathered in February 2019 in Winnipeg, Canada, for a workshop titled 'Breastfeeding and the Origins of Health: Interdisciplinary Perspectives and Priorities'. The mandate of this workshop was to identify and discuss research priorities and anticipated breakthroughs in breastfeeding or HM research (Section 2); discuss the perceived barriers and challenges to achieving these breakthroughs (Section 3); and outline a plan of action towards supporting and maximizing the impact of future breastfeeding and HM research (Section 4). Participants were invited on the basis of their expertise in breastfeeding and HM research or practice, with consideration for equity, diversity and inclusion across disciplines, settings and career stages. Not everyone who was invited was able to attend, and a few declined participation precisely because of the

inequities and barriers
Breastfeeding is among the most cost-effective public health interventions available, providing protection against several short-and longterm health conditions for both mother and infant , which reduces healthcare costs (Rollins et al., 2016). The WHO recommends that all infants be exclusively breastfed for around 6 months and continue breastfeeding with complementary foods until 2 years or beyond (WHO, 2003), yet by 6 months of age, only 58% of US infants are breastfed and just 25% are exclusively breastfed (Centers for Disease Control and Prevention, 2018). Rates are lower in the United Kingdom (34% any breastfeeding at 6 months), the Netherlands (32%) and France (23%) . In many settings, breastfeeding rates are even lower among infants born to minority and/or low-income mothers, which may contribute to long-term health inequities in these marginalized populations (Anstey, Chen, Elam-Evans, & Perrine, 2017;Merewood et al., 2019;Patel et al., 2019).
Barriers to breastfeeding include stigma, lack of support and structural factors that disproportionately affect marginalized populations (e.g., lack of breastfeeding education and support services and inadequate maternity leave policies) (Nickel et al., 2014). Social determinants of health and cultural factors also influence breastfeeding outcomes (Byrd, Balcazar, & Hummer, 2001;Cattaneo, 2011;Celi, Rich-Edwards, Richardson, Kleinman, & Gillman, 2005;Dubois & Girard, 2003;Patel et al., 2019). Of great concern, the breastfeeding gap within populations is widening (Li et al., 2019;Logan et al., 2016;Nickel et al., 2014). It is critical to understand the reasons for this disparity and to collaboratively develop context-specific strategies to address them.

| Implementation science
Addressing low breastfeeding rates and breastfeeding inequities requires implementation science (Pérez-Escamilla & Hall Moran, 2016) to translate research into evidence-based advocacy efforts, policies and large-scale programmes. Implementation science involves mixedmethods approaches to design, evaluate and scale up effective programme innovations, and strategies to enhance the use of existing knowledge, tools and frameworks based on a systems thinking approach (Tumilowicz et al., 2019). Coordinated efforts by multidisciplinary teams are required to execute planning, collaboration, monitoring and adjustments. Implementation science has been applied successfully to scale up effective breastfeeding programmes across world regions using the breastfeeding gear model (Pérez-Escamilla, Curry, Minhas, Taylor, & Bradley, 2012) and building upon evidencebased interventions (Merewood et al., 2019;Nickel, Taylor, Labbok, Weiner, & Williamson, 2013; 2.2 | Improving awareness of evidence-based benefits, challenges and complexities of breastfeeding among health practitioners and the public using effective messaging platforms 2.2.1 | Lack of awareness and competing/ inconsistent messaging Evidence-based and culturally competent engagement about breastfeeding remains a constant challenge, particularly when contrasted by the sophisticated messaging strategies used by infant formula companies (Seals Allers, 2018). This challenge is compounded by a lack of formal education about lactation and breastfeeding support for most healthcare professionals (Freed et al., 1995;Younger Meek, 2019). There is also a lack of rigorous science investigating the implications of breastfeeding and/or HM on infant health. At the same time, health-focused research and messaging often fail to acknowledge that many women want to breastfeed for cultural or religious reasons, or simply because it is a physiological norm and a reproductive right, regardless of any health benefits (Brown, 2018). This constellation of challenges has resulted in public confusion and inconsistent messaging regarding breastfeeding and HM.

| Reaching everyone with appropriate messaging
Supporting breastfeeding is a societal responsibility (Rollins et al., 2016). Mothers and infants are underserved by societies that deprive families of the autonomy and information to make evidencebased decisions about infant feeding, invalidate mothers' emotions and desires to breastfeed (or not), default to infant formula rather than effectively supporting breastfeeding and undervalue the time and energy that women dedicate to breastfeeding (Brown, 2018).
Messages should not focus on the individual mother alone; they should be adapted for traction across all stakeholders that influence breastfeeding success-from grandparents and clinicians to employers, business owners and political bodies. It is also important to 'normalize' breastfeeding for the next generation of families through embedding breastfeeding education in school curriculums (Glaser, Roberts, Grosskopf, & Basch, 2015).
Messages must be culturally sensitive and recognize that, in some countries, inequities in breastfeeding have resulted from historical trauma and discrimination against marginalized communities (Asiodu & Flaskerud, 2011;Heart, Chase, Elkins, & Altschul, 2011).
Effective initiatives built within these communities are foundational models for achieving inclusive care (e.g., Momma's Village, Indigenous Breastfeeding Counsellor, and Reaching Our Sisters Everywhere: African American Breastfeeding Blueprint) (Bugg & Bugg, 2013).
Healthcare providers (Pound, Moreau, Hart, Ward, & Plint, 2015) and policymakers must be properly and comprehensively trained, as messaging to promote breastfeeding will have limited success without equitable policies that support and protect breastfeeding at the individual, institutional and societal levels.

| Leveraging social media and online communities
Social media platforms provide a global medium to amplify public health campaigns, influence health behaviours and establish social norm (Giustini, Ali, Fraser, & Kamel Boulos, 2018;Merchant, 2020).
Social media can be used to share educational and supportive messaging about breastfeeding and HM (Marcon, Bieber, & Azad, 2018;Price et al., 2018); however, it can also facilitate dissemination of pseudoscience and provide a platform for divisive agents (Giustini et al., 2018). Opportunities exist to spread breastfeeding messaging more broadly and effectively using social media (Brown, 2016), smartphone apps (Coughlin, 2016), animations (e.g., bit.ly/2euMoxh), interactive infographics (e.g., human-milk.com) and popular science writing. Academics studying breastfeeding and HM should make better use of these 'nontraditional' forms of knowledge translation or actively engage with messaging experts to maximize the reach and impact of their research. Feeding bottled HM may not be biologically equivalent to feeding at the breast. Differences have been observed for infant weight gain , satiety (Li, Fein, & Grummer-Strawn, 2010), asthma (Klopp et al., 2017) and memory (Pang et al., 2019), suggesting a potential negative impact from the process of bottle feeding and/or reduced bioactivity of expressed HM. However, feeding expressed HM still provides benefits compared with infant formula Klopp et al., 2017) and should be encouraged when nursing is not possible or preferred. Future research should capture the complexity of modern HM feeding practices, even among exclusively breast (milk)-fed infants. As new evidence emerges, guidelines (Eglash & Simon, 2017) may require revision to provide up-to-date advice for storing and feeding expressed HM. It is also critical to address the structural barriers that force women to choose between pumping and stopping breastfeeding altogether.

| Donor milk and milk sharing
The availability and use of donor HM (DHM) is increasing. In preterm infants, access to DHM (as compared with infant formula) lowers the risk of developing necrotizing enterocolitis (Quigley, Embleton, & McGuire, 2019) and can support the establishment of the mother's own milk supply (Kantorowska et al., 2016;Wilson et al., 2018) but may result in lower growth rates (Quigley et al., 2019). Research is needed to identify best practices, including whether and how pooling (Young et al., 2018) and pasteurizing (Ewaschuk, Unger, Harvey, O'Connor, & Field, 2011) should be conducted to preserve the bioactive integrity. 'Personalizing' DHM is another area requiring innovation-for example, by matching DHM on maternal and/or infant characteristics or using mother's own milk to seed the microbiota of DHM (Cacho et al., 2017). Research is also needed to inform prioritization of DHM allocation and improve milk banking processes (Matthews et al., 2019).
The limited access to DHM in most countries has led to a large increase in unregulated informal HM sharing (Palmquist et al., 2019). To prevent potential harms from these practices, a pragmatic approach has been proposed by the Academy of Overall, modern caregivers are actively seeking practical advice (Lupton, 2016) to inform their diverse feeding regimens, and much more research is necessary to provide evidence-based recommendations. This will require researchers to explore and document alternative feeding modes and engage with diverse stakeholders including breast pump manufacturers, donor milk banks and regulatory agencies. suggesting that women can be characterized according to their milk composition profile (Munblit et al., 2017) and that variation in combinations of milk components rather than single factors may be linked with infant health.

| Determinants of human milk composition
Many HM constituents vary greatly between and within populations,  (Fang et al., 2017) and are warranted to examine HM composition.
Geographic variation in HM composition has also been described (Gay et al., 2018;Kumar et al., 2016;McGuire, Meehan, Brooker, et al., 2017;Munblit et al., 2016;Ruiz et al., 2017) Waidyatillake et al., 2018). This is problematic because when claims are publicly refuted, trust in the scientists and health professionals producing and conveying these claims could be eroded, potentially leading to a backlash against researchers and breastfeeding promotion efforts. Robust evidence quantifying specific health effects (or lack thereof) and their mechanisms will be key to producing reliable cost-benefit analyses and advocating for more investment in services to protect, promote and support breastfeeding.

| Epidemiological studies of breastfeeding
It is not ethical to randomize breastfeeding, so almost all evidence supporting or refuting breastfeeding or HM feeding comes from observational studies or animal models. Epidemiological studies vary in design, size and setting (e.g., low-vs. high-income countries), and their collective results reflect considerable heterogeneity for many of the outcomes studied . Heterogeneous

| Negative personal experiences with breastfeeding can fuel 'breastfeeding denialism' and impede research progress and translation
Public discussions about infant feeding in mainstream and social media highlight the deeply personal nature of infant feeding experiences. Women express the joy they experience while breastfeeding and share their struggles and emotional turmoil when they are unable to meet their own breastfeeding goals (Brown, 2018). Negative or 'denialist' attitudes towards breastfeeding are sometimes fuelled by individuals with negative personal experiences (Palmer, 2019), which often originate from disempowering interactions with healthcare systems (Brown, 2018). Researchers face complex challenges when discussing the health benefits of breastfeeding because, although

Box 1 Sources of heterogeneity in epidemiological studies of breastfeeding
A) Bias-inducing limitations (measuring the same effects, but with bias) • Confounding-The main barrier to inferring causality in observational studies of breastfeeding is confounding by socioeconomic factors and variables following a socioeconomic gradient (e.g. maternal health and lifestyle). This is because establishing or continuing to breastfeed is strongly associated with socioeconomic status (particularly in high-income countries), as are many of the health outcomes.
• Selection bias-Studies focused on determinants of breastfeeding duration can be biased if they exclude mothers who do not initiate breastfeeding, particularly if these same determinants also affect initiation (Paternoster, Tilling, & Davey Smith, 2017).
• Publication bias-Compared to studies suggesting no relationship between breastfeeding and a health outcome, those showing positive associations are more likely to be published (Horta & Victora, 2013), thus affecting conclusions drawn in systematic reviews and meta-analyses.

B) Non-bias inducing limitations (measuring different effects)
• Misclassification of breastfeeding exposures-Standardized definitions have been proposed for breastfeeding research, but many studies do not apply them (Miliku & Azad, 2018). Ideally, studies should capture and distinguish the following:

Duration and exclusivity of breastfeeding
Nursing at the breast vs expressed HM (relative proportion of each; storage of expressed milk)

Perinatal feeding exposures in hospital
Introduction of complementary foods (both age and type/quality of food)

If partially breastfed: relative proportion of HM vs infant formula
If bottle fed (whether infant formula or HM): feeding style If formula fed: variation in type of infant formula used (e.g. high/low protein, protein source and size, percentage carbohydrate from lactose, addition of pre/probiotics, lactoferrin, milk fat globule membrane, etc.) • Failure to address effect modifiers and interactions-There may be genuine differences in breastfeeding effects when breastfeeding interacts with setting-specific cultural/environmental factors. Such interactions are rarely addressed but should be considered. Possible modifiers include the following (though it should be noted that some of these factors could also be confounders).
Maternal diet, lifestyle and drug use (prescription or recreational)

Maternal physical and mental health
Maternal/parental attachment and parenting style Environmental exposures that are mitigated or exacerbated by breastfeeding (e.g. pollution, smoking)

Differences in HM composition (see Section 2.3)
advancing research on this topic will ultimately improve health for all mothers and infants, it also perpetuates a dialogue that can cause guilt among women who did not breastfeed. These personal biases can impede research progress and impact by influencing the peer review process and the translation of research results. One way to address this challenge is to avoid focusing entirely on the mother-infant dyad and their (in)ability to breastfeed, which ignores the myriad underlying social and structural determinants that affect this process, as discussed in Section 2.2. Another way to address this challenge is to undertake qualitative research focused on understanding the lived experiences of families who have struggled with breastfeeding (Spencer, 2008). It is also important that women unable to breastfeed are supported through research on alternative feeding methods and responsive bottle feeding.

| Distinguishing and aligning advocacy and research efforts
Advocacy is central to advancing public health agendas, including breastfeeding (Michaud-Létourneau, Gayard, & Pelletier, 2019;Pérez-Escamilla et al., 2012;Rosen, 1993). In parallel, marketing messaging by industry has been used heavily to advocate for infant formula (Robinson, Buccini, Curry, & Pérez-Escamilla, 2018 Claims may also be unsubstantiated in breastfeeding advocacy and promotion efforts. There is a danger that interesting new clinical or laboratory findings related to HM (e.g., the presence of stem cells) may be used prematurely by advocacy groups before the direct benefits to infant or maternal health are understood. Such claims can inadvertently undermine the support of breastfeeding by implying that further research is not needed and giving the impression that research in this field is not sufficiently rigorous. Such claims may later be used by industry to justify adding new ingredients to infant formula without appropriate evidence.

| Supporting advocacy with evidence
As a general strategy, population-wide efforts to improve science

| COIs in breastfeeding and human milk research
COIs in research put the process at risk by potentially biasing a researcher's professional judgement (Suter & Cormier, 2015). COI can emerge across a variety of dimensions when an individual has a personal, professional or financial interest that could affect how they carry out or interpret their work. Here we focus on financial COI and offer a discussion of the challenges and opportunities afforded by working with industry partners for breastfeeding and HM research. Workshop discussions highlighted the diversity of opinions on this topic.
3.4.1 | Industry sponsorship of breastfeeding and human milk research can lead to bias and incorrect public health messaging Over the past century, many industries have funded research as a strategy for gaining public credibility and acquiring market share (Bekelman, Li, & Gross, 2003;Flacco et al., 2015;Lexchin, Bero, Djulbegovic, & Clark, 2003;Lundh, Lexchin, Mintzes, Schroll, & Bero, 2018). This practice extends to breastfeeding and HM science, where the infant feeding industry invests heavily in breastfeeding and HM research (Shenker, 2018;Van Tulleken, 2018). Industry funding may influence decision making in academic healthcare settings on an unconscious level, reflecting 'motivational bias' (Dana & Loewenstein, 2003). The act of declaring COI may actually exaggerate rather than mitigate this form of bias (Cain, Loewenstein, & Moore, 2005. Scientists who ignore the risks of motivational bias can inadvertently facilitate the dissemination of incorrect public health messages (Bekelman et al., 2003;Campbell, Louis, & Blumenthal, 1998;Smith, 2006;Thompson, 1993). There is no guarantee that open declarations of COI will prevent such bias when researchers accept grant funding (whether or not it is restricted) from companies with vested interests in the outcomes generated. Further, a randomized study (Sharek, Schoen, & Loewenstein, 2012)  In particular: • To address breastfeeding inequities experienced by marginalized communities, mixed-methods implementation research is needed to engage families and codevelop context-specific solutions, followed by cost-effective scale-up of effective policies and programmes.
• To improve awareness about breastfeeding among health practitioners and the public, and support evidence-informed advocacy efforts, researchers should develop and adapt messaging for diverse stakeholders.
F I G U R E 3 Key priorities and anticipated breakthroughs, barriers and challenges, and recommendations for research on breastfeeding and human milk (HM) • To generate much needed knowledge about alternative methods of HM feeding (e.g., pumping, donor milk and milk sharing), researchers should accurately capture feeding practices.
• To evaluate the causal health effects of breastfeeding and HM, studies should be rigorously designed, carried out, analysed and interpreted to mitigate bias.
• To advance our knowledge of HM composition, synthesis and consumption, it is essential to apply standardized and validated sampling and analytical methods, to evaluate milk as a whole instead of a mixture of discrete components and to measure the volume of milk consumed accurately.
• To support evidence-informed advocacy efforts, researchers should provide clear evidence summaries of their findings, discredit unsubstantiated claims and actively participate in initiatives leading to effective public policy and advocacy recommendations.

| CONCLUSION
Breastfeeding and HM research is vital to understanding and improving health worldwide. As summarized in Figure 3, this transdisciplinary field is on the cusp of major discoveries with implications for lifelong health. However, unlike many other areas of health research, this field is laden with emotion and denialism. It is also challenged with informing yet remaining distinct, to some extent, from breastfeeding advocacy efforts. To advance research in this field and maximize its reach and impact, larger research investments are needed and interdisciplinary collaboration is essential; the scientific community must properly manage COI and engage families and other stakeholders in research planning and knowledge translation efforts.

CONFLICTS OF INTEREST
The authors have declared all relevant potential sources of conflict of interest, including salary/research funding, patents, stock ownership, speaking activities at sponsored conferences/workshops/events, consulting, boards, societies, committees, expert panels and community organizations, detailed in Table S1.