Testing the buffering hypothesis: Breastfeeding problems, cessation, and social support in the UK

Testing the buffering hypothesis: Breastfeeding problems, cessation, and social support in the UK. Abigail E. Page, Emily H. Emmott, Sarah Myers. American Journal of Human Biology, May 30 2021. https://doi.org/10.1002/ajhb.23621

Abstract

Objectives: Physical breastfeeding problems can lead women to terminate breastfeeding earlier than planned. In high-income countries such as the UK, breastfeeding problems have been attributed to the cultural and individual “inexperience” of breastfeeding, ultimately leading to lower breastfeeding rates. Yet, cross-cultural evidence suggests breastfeeding problems still occur in contexts where breastfeeding is common, prolonged, and seen publicly. This suggests breastfeeding problems are not unusual and do not necessarily lead to breastfeeding cessation. As humans evolved to raise children cooperatively, what matters for breastfeeding continuation may be the availability of social support during the postnatal period. Here, we test the hypothesis that social support buffers mothers from the negative impact breastfeeding problems have on duration.

Methods: We run Cox models on a sample of 565 UK mothers who completed a retrospective online survey about infant feeding and social support in 2017–2018.

Results: Breastfeeding problems were important predictors of cessation; however, the direction of the effect was dependent on the problem type and type of support from a range of supporters. Helpful support for discomfort issues (blocked ducts, too much milk) was significantly associated with reduced hazards of cessation, as predicted. However, helpful support for reported milk insufficiency was assoicated with an increased hazard of cessation.

Conclusions: Experiencing breastfeeding problems is the norm, but its impact may be mitigated via social support. Working from an interdisciplinary approach, our results highlight that a wide range of supporters who provide different types of support have potential to influence maternal breastfeeding experience.

5 DISCUSSION

In our sample of 565 relatively affluent and well-educated women from the UK, we find that almost everyone reported problems breastfeeding, underlining that breastfeeding problems are the norm rather than the exception. Such problems potentially explain why, as reported elsewhere, 71% of respondents who stopped breastfeeding prior to 8 weeks found infant feeding stressful and 60% emotionally draining (Myers et al., 2021). As women often report feeling unprepared, abnormal and isolated when they encounter common issues (Brown, 2016; Wall, 2001), every attempt should be made to inform women antenatally that breastfeeding is a learnt behavior (Varki et al., 2008) and how best to prepare for, and overcome, future challenges (Brown, 2016; Emmott et al., 2020b).

The women in this study were, overall, well-supported. Unhelpful support was rarely reported, and women frequently indicated a wide range of supporters as providing helpful support. Partners were documented as particularly helpful, as were midwives and health visitors. Beyond these well-established supporters, we also see that the mother's mothers (infant's maternal grandmothers), friends and sisters were also providing helpful support across informational, practical and emotional domains. This indicates that multidimensional social support is an important feature of the postnatal period in the UK, even though women, especially those who are more highly educated, frequently live further from family and friends (Chan & Ermisch, 2015). As we have argued previously (Emmott et al., 2020a, 2021; Emmott & Mace, 2015; Myers et al., 2021), the benefit of an evolutionary anthropological approach is the exploration of investment transfers to mothers from a wide range of supporters due to the emphasis on cooperative childrearing (Emmott & Page, 2019). An evolutionary framework highlights that there are many mechanisms by which support can work, and support which can facilitate breastfeeding need not be limited to health care professionals.

5.1 Hypothesis 1: Breastfeeding problems increase the likelihood of cessation

While breastfeeding problems are frequently reported as reasons to stop breastfeeding (Dewey et al., 2003; DiGirolamo et al., 2005; Verronen, 1982), our results highlight that different types of problems have different associations with breastfeeding cessation. As is demonstrated consistently in other studies (Ahluwalia et al., 2005; Ingram et al., 2002; Kirkland & Fein, 2003; Lamontagne et al., 2008), we find that women who reported not having enough breast milk, or struggling to obtain a good latch, were much more likely to terminate breastfeeding within 3 months. These two problems can be understood as relating to infants' nutritional intake (Ingram et al., 2002; Kirkland & Fein, 2003; Li et al., 2008; Verronen, 1982), as a poor latch can result in less efficient feeding. Some have suggested that insufficient milk is often provided as a reason for stopping breastfeeding because the focus on the infant's wellbeing is a more “acceptable” reason to terminate breastfeeding within a society which often relates breastfeeding to “good” mothering (Ingram et al., 2002; Whelan & Lupton, 1998). While we have no data that can directly speak to whether insufficient milk was overreported, it is worth noting it was the second least reported problem. We also did not ask women why they stopped breastfeeding, thus reducing the motivation to inaccurately report and increasing confidence that insufficient milk was perceived to be an actual problem by participants. In contrast, we found that women who reported having problems with blocked ducts, mastitis, sore nipples or too much breast milk were less likely to terminate breastfeeding. These problems can be conceptualized as relating more to mothers' (not insignificant) pain and discomfort, or issues which require management (i.e., pumping excess milk) (Binns & Scott, 2002). Some women may be more able to endure issues of breastfeeding when it is at their own cost, rather than their infant's, which may relate to common parenting ideals focused on intensive mothering—a primarily white, middle-class emphasis on child centered and self-sacrificial parenting (Reyes-Foster & Carter, 2018). Therefore, as argued by Fahy and Holschier (1988), successful breastfeeding occurs not in the absence of problems, but in the mother's ability to overcome these problems (Binns & Scott, 2002).

5.2 Hypothesis 2: Social support will moderate the negative relationship between breastfeeding problems and cessation

In the un-moderated model, blocked ducts and too much breast milk were associated with reduced hazards of breastfeeding termination. However, by exploring the interactions between social support and breastfeeding problems, it was evident that this effect was driven by support that was considered helpful. Across the blocked ducts models, the lowest hazards of stopping breastfeeding were associated with helpful practical and informational support. Furthermore, women who had issues around blocked ducts or too much breast milk and reported unhelpful informational support (from friends, partners and midwives) were more likely to stop breastfeeding prior to 3 months. It has been reported elsewhere that inconsistent conflicting advice from HCPs results in mothers becoming frustrated, confused and more likely to cease breastfeeding (Garner et al., 2016; Ingram et al., 2002; Lamontagne et al., 2008). While we cannot speak to the specifics of the informational support received by our participants, our results nonetheless indicate that this effect may go beyond HCPs, with unhelpful advice from partners and friends having similar effects.

Contra to predictions, rather than reducing the negative impact of milk insufficiency on duration, support of all types—but particularly emotional support—from sisters, health visitors and midwives was associated with increased likelihoods of cessation. While the direction of causality is unknown, this relationship may be related to a woman's support needs. After experiencing perceived milk insufficiency mothers may wish to switch to formula to ensure infant's weight gain. Thus, informational and practical support may help them wean their infants. Supporters who empathize with a mother's situation or affirm her decision to stop breastfeeding may be experienced as being emotionally supportive. Levels of breastfeeding intent in our sample were high (Myers et al., 2021), potentially increasing the importance of emotional support for women who decide to stop early. This may be particularly important in a UK context where breastfeeding is currently heavily promoted, thus to cease breastfeeding is to go against a central tenant of the prevailing, White middle-class mothering discourse (Crossley, 2009; Faircloth, 2015; Kukla, 2006; Lee, 2007). While our study did not explore women's mental wellbeing, the relationship between breastfeeding expectations, problems and postnatal depression has been well documented (Brown et al., 2016; Shakespeare et al., 2004). This highlights that social support during the postnatal period is not only about prolonging breastfeeding but also about supporting mothers mental wellbeing (Emmott et al., 2020b; Trickey, 2018; UNICEF, 2018).

As indicated above, emotional support goes beyond friends and family. In our interaction models, emotional support moderated the relationship between various problems and duration when it originated from all types of health professionals. Graffy and Taylor (2005), in a qualitative analysis of interviews with 654 women from the UK, report that although women requested informational support, women were more often sensitive to the way HCPs treated them. Alongside practical tips and guidance, they wanted acknowledgement of their experiences and to be reassured that issues during breastfeeding were normal, and thus were encouraged to continue. Clearly, emotional support from HCPs has an important role to play. This is highlighted by the negative impact that poor emotional support from GPs had in our study. Emotionally unsupportive GPs could be increasing maternal stress, making breastfeeding more difficult and further adding to mothers' concerns. Conversely, emotional support, through acknowledgement, reassurance and encouragement (Graffy & Taylor, 2005) may increase a mother's self-efficacy to deal with latching, or other issues.

Support from a wide range of sources interacted with breastfeeding problems to predict duration. However, no significant effects were found for the mother's mother, father and brothers. This result is perhaps less surprising for brothers, given their lower levels of helpful support apparent in Figure 1, but much more surprising for mothers' mothers as they have been identified as key supporters of women in the postpartum period and beyond (Scelza, 2009; Scelza & Hinde, 2019; Sear & Coall, 2011; Sear & Mace, 2008; Snopkowski & Sear, 2015). It may be that our sample lacks the variance required to explore our question in relation to mother's mothers, as the vast majority of women reported helpful support from them. Further analyses in a more diverse sample may help unpack this. Peer supporters also demonstrated a different trend, in which the hazard of termination was the lowest when their support was absent. This probably stems from the fact that unlike GPs, health visitors and midwives, not everyone will encounter a peer supporter (as indicated in Figure 1) and likely only do so when they are facing considerable problems. Therefore, those people receiving help from peer supporters may already be more likely to stop breastfeeding.

5.3 Hypothesis 3: The type of support impacts the moderating effect

Practical support is framed within evolutionary approaches as reducing, or having the potential to reduce, a mother's workload, meaning mothers will have more energy to invest in tasks which maximize lifetime fitness, as construed in the current environment (Emmott & Page, 2019; Kramer & Veile, 2018; Meehan et al., 2013; Page et al., 2021). Consequently, while not true of all types of practical support (e.g., allofeeding—individuals other than the mother feeding the infant [Emmott & Mace, 2015; Myers et al., 2021]), overall, practical support is hypothesized to increase breastfeeding duration. In our sample, in which the majority expressed a wish to breastfeed (Myers et al., 2021), mothers who received helpful practical support from family members may have been able to focus on breastfeeding. This additional energy devoted to breastfeeding may have allowed them to work through problems, providing the time required to access specialist breastfeeding support.

Interestingly, in our sample, moderating practical support was received from the partner's parents (the infant's paternal grandparents), rather than the mothers' parents (the infant's maternal grandparents). Who helps mothers is likely to be partially context specific; nonetheless, evolutionary theories of kinship do predict differential investment by paternal and maternal grandparents (Beise, 2005; Gibson & Mace, 2005; Sear, 2008; Sear & Coall, 2011), suggesting that relatives from different lineages may invest in different types of supportive activities (Sear & Mace, 2008). However, it may also be the case that, as noted above, there is simply more variance in partner's family for us to pick up on these trends. It is also important to note that, by focusing on helpfulness, in this analysis we have explored all types of practical support collectively, combining allofeeding with other forms of practical support. Since allofeeding has been demonstrated to have a negative relationship with breastfeeding duration (Emmott & Mace, 2015), this may be confounding results in relation to supporters most likely to perform allofeeding—which in this sample are the partner and mother's mother (Myers et al., 2021).

Our data indicate that emotional and practical support were important moderators; however, contra predictions, so was informational support. While this hypothesis was not formally designed to separate the independent effects of emotional, practical and informational support, we did see that informational support was the most frequent moderator, and the receipt of helpful informational support was often associated with a lower likelihood of breastfeeding cessation. This underlines the fact that breastfeeding is a learnt behavior (Volk, 2009) and suggests while women benefit from practical and emotional support helping them to persist in spite of problems, informational support may curtail the duration of the problem itself. It may also be that the usefulness of informational support is dependent on its delivery alongside emotional support (Fallon et al., 2017; Fallon et al., 2019; Trickey, 2018). Future work should tease out these effects in greater depth.

Informational support was not limited to HCP but also received from partners, sisters and friends. Similar results have been found elsewhere; for example, Swedish mothers whose own mothers had discussed breastfeeding with them were more likely to breastfeed for longer, reporting greater increased confidence (Ekström et al., 2003). This demonstrates the importance of information and advice from a range of supporters. Consequently, researchers and public health specialists need to consider where else information is coming from and how to direct (helpful) information-based interventions beyond the mother (Daniele et al., 2018; Negin et al., 2016; Wolfberg et al., 2004).

Our focus on the physiological experience of breastfeeding problems and the individual experience of social support is not to suggest that wider socioeconomic and political factors are not key predictors of breastfeeding, nor is it easy to untangle the impact of behavior, norms and structural factors (Palmquist & Doehler, 2014). There are large inequalities in infant feeding experience along structural lines in the UK and similar HIC (Victora et al., 2016), contributing to socioeconomic gradients in inflammation and infant weight (McDade & Koning, 2021). These inequalities exist due to cultural and religious norms around breastfeeding, particularly in public (Chang et al., 2021), access to social support (Grubesic & Durbin, 2020; Tomori, 2009), opportunity costs of breastfeeding (Hough et al., 2018; Tully & Ball, 2018), as well having convenient and quick-to-access places to breastfeed (Brown et al., 2020; Hauck et al., 2020). Furthermore, the experience of breastfeeding problems is unlikely to be evenly distributed as one study found that young, unmarried and non-college educated US women were more likely to experience breastfeeding problems resulting in disrupted lactation (Stuebe et al., 2014). Therefore, our results may be underplaying the importance of breastfeeding problems given our sample of educated, affluent white women, who likely have privileged access to formal and informal social support. Here, we have demonstrated the importance of variation in social support in moderating the relationship between breastfeeding problems and duration, but further research with a more diverse sample is required to explore what causes variations in this support.

5.4 Limitations

We have already noted that our sample is a key limitation in this research. While 565 women is not an insignificant sample size, there was a lack of diversity in breastfeeding durations and support received. This is a product of the homogenous nature of our sample, which is largely educated, affluent and white—a clear limitation of this study. As a result, statistical power was likely an issue in our models increasing the likelihood of Type II errors. For this reason, while we often see nonoverlapping confidence intervals between helpful and unhelpful support, the intervals for unhelpful support are often wide and spanning one making interpretation difficult. Our sample was recruited online using convenience-sampling, which likely biased it to more affluent women (Topolovec-Vranic & Natarajan, 2016). This issue is not uncommon within survey-based breastfeeding studies and should be addressed in future research. Middle-class, more affluent women have the time, energy and desire to engage with scientific studies; more needs to be done to make this process low cost and desirable to a wider demographic. An additional concern with online-based data collection are programs which automatically fill in surveys (e.g., bots) (Dupuis et al., 2019) and low effort respondents, both which are likely to occur when financial incentive is offered for survey completion (Buchanan & Scofield, 2018). No financial incentive was offered for the present survey, reducing our concern regarding bots—which is further diminished by the absence of suspiciously rapid completion times. Furthermore, the majority (78.6%) of respondents invested significant effort into responding to a number of optional open-text questions (not used in this analysis, but utilized in Emmott et al. (2020a))—these would be expected to be skipped by low-effort respondents and either skipped, answered incoherently, or repetitively by bots, which data exploration points against.

The second limitation is the potential for reporting bias due to the retrospective design of this study. We asked women with children aged up to 24 months of age about the problems they experienced and support they received since giving birth. Women may forget key early events in the light of later ones, and the perceived severity of a problem is likely impacted by the severity of later ones (Williamson et al., 2012). Further, given the retrospective nature of this study we have captured breastfeeding problems in a simplistic fashion. Our binary measure of yes/no hides variation in severity, duration, and number of occurrences as well as varying causes or exacerbating factors (such as maternal or infant factors). Prospective study designs are better-suited to explore the causal relationship between support, problems and breastfeeding—all factors which fluctuate on a daily basis—and we encourage their future use.