Breastmilk has been widely advocated as the optimal nutrition for newborn babies and infants. In both developing and developed countries, extensive research has produced evidence to sustain claims that early nutrition has a direct influence on infant morbidity and mortality (Department of Health, 2014), and that, for most infants, breastmilk has considerable advantages over formula substitutes. The World Health Organization (WHO) estimates that, globally, more than 800 000 children under the age of 5 years old could have been saved every year if optimally breastfed (WHO, 2014). Despite this, the UK, a developed country, is placed among those with the lowest breastfeeding rates in the world.
Recommendations advise exclusive breastfeeding for the first 6 months of an infant's life (Department of Health, 2003) but the latest figures show that, although 72.6% of women initiate exclusive breastfeeding at birth (NHS England, 2017), prevalence dramatically drops to 30% at 6–8 weeks after birth (Public Health England, 2017) and to only about 1% by the time the infant is 6 months old (McAndrew, et al, 2012).
UK views on breastfeeding
Across the UK, the popularity of infant formula grew in the first half of the 20th century and, by the 1970s, the Government started to become concerned with the impact that this trend was having on social norms, morbidity, inequalities in health and, consequently, costs (McAndrew, et al, 2012). Since then, governments have considered low breastfeeding rates a public health issue, and have attempted to raise them through various programmes—although with little success.
Historically, breastfeeding promotion in the UK has been made mostly through strong advertisement that ‘Breast is Best’ and infant formula, as a general rule, labelled as ‘close second best’. The campaign for breastfeeding relies heavily on promotion and the view that, although the infant feeding method that a mother adopts is a choice, all women and babies should be able to breastfeed successfully. The campaign does not take into account the difficulties that many mothers experience (Lee and Furedi, 2008), and the fact that sometimes breastfeeding is not an informed choice. Breastfeeding support can be a ‘postcode lottery’, with some areas having full Baby Friendly Initiative accreditation and others only a certificate of commitment (Public Health England and UNICEF, 2016). This puts mothers in very different positions when it comes to making an informed choice, support for their needs and protection of their rights. In theory, all mothers can breastfeed, but this only happens when they are well informed, well cared for by the health system, and well supported by their family and wider society (WHO, 2017).
Breastfeeding policy and strategy
Underpinned by a robust body of evidence, in 2016, concerning allegations were released of the significant global economic and health benefits that governments can generate when they make a political commitment to supporting breastfeeding (Rollins, et al, 2016). In this global publication on breastfeeding, the UK was reported as having the lowest breastfeeding rates in the world at 12 months post birth (Victora et al, 2016). In reaction to this, and in the absence of a strategy or clear plan, Public Health England, supported by the UNICEF UK Baby Friendly Initiative created ‘Infant Feeding Commissioning Services’, a toolkit for local authorities to improve breastfeeding rates. But, once again, the document lacked target, specific objectives, timeframes, accountability and a plan for evaluation—all essential characteristics in good policy making (Hallsworth and Rutter, 2011).
This report generated movement, however. Health professionals, lactation experts, academic researchers and professional organisations co-ordinated and, in an open letter to the Government, called for action to support mothers to breastfeed. Meanwhile, the first World Breastfeeding Trends Initiative (WBTi) report was also presented. The report identified the gaps and weaknesses in evidence-based key indicators that promote and sustain breastfeeding (WBTi Steering and Core Groups, 2016). England scored particularly low on government commitment initiatives such as national policy or strategy, information support, monitoring and evaluation of data (WBTi Steering and Core Groups, 2016).
Barriers to breastfeeding data
Once again, the Government recognises the low breastfeeding rates as a public health concern but the lack of commitment to tackle it is evident and incomprehensible. Examples of this are the cessation of the Infant Feeding Survey in 2015 and the massive health and social care budget cuts in the last 10 years, which have threatened breastfeeding support services. In the absence of a dedicated strategy to breastfeeding, the funding for its promotion and support is under the umbrella of every health and social care budget. Analysing the data since 2010, the initiation rates for breastfeeding dropped from 73.7% to 72.6% (NHS England, 2017) and rates of continuation at 6–8 weeks from 46.2% (Department of Health, 2011) to 44.4% (Public Health England, 2017). In the absence of the Infant Feeding Survey, breastfeeding rates at 4, 6 and 12 months are not being reported, making clear analysis difficult. In addition, other important information, such as reasons why women discontinued breastfeeding and what their initial feeding intentions were, stopped being collected. The last Infant Feeding Survey found that 86% of the mothers who stopped breastfeeding in the first 2 weeks wanted to breastfeed for longer. This is a very high rate, which requires consideration.
Determinants of health behaviour
To analyse trends in health behaviours, such as what makes women decide to bottle-feed exclusively from the moment of birth or to give up breastfeeding in the first few weeks, theorists have used several approaches to explain health behaviours and beliefs. Early studies found that people's reactions to illness were dependent on their economic and cultural situation, and that decisions to seek professional help were mediated by social factors (Nettleton, 2006). The social constructionist approach assumes that a phenomenon is:
‘The result of historical, social and political processes, rather than an inevitable result of our greater understanding of the “reality” of the body, or disease.’
Research later shifted to understanding health beliefs and to developing complex and sophisticated theories about the maintenance of health (Nettleton, 2006). The Health Belief Model is the earliest and most widely used. The model theorises that health behaviour is determined by four factors: perceived susceptibility to a condition, perceived severity of the condition, perceived benefits of actions to prevent the condition, and perceived barriers or costs of the action (Coreil, 2009). It therefore assumes that the decision to act is the result of an assessment of the benefits, available alternatives, efficacy and perceived costs or barriers of that behaviour (Conner, 2010).
Several research studies (Celi et al, 2005; Al-Sahab et al, 2010; Ahluwalia et al, 2012; McAndrew et al, 2012) have examined the relationships between different measures of sociodemographic, economic and attitude characteristics and breastfeeding, however, little attention has been given directly to determinants of bottle-feeding. Taking into account this research gap, this study aimed to determine predictors of bottle-feeding by examining socioeconomic, demographic, family-related, and birth factors, while adjusting for the remaining variables. By examining the relationships between the factors that predict bottle-feeding, this study also investigated differences between the mothers who bottle-fed from birth and those who attempted breastfeeding but stopped early in their baby's life. The research questions used were the following:
Methodology
Design
This research study is a secondary analysis of the Infant Feeding Survey 2010 (McAndrew et al, 2012), a nationally representative UK dataset on infant feeding practices published by the Health and Social Care Information Centre that has been subject to ethical review.
Setting
The Infant Feeding Survey was a large-scale quantitative dataset that collected the estimates of the incidence, prevalence and duration of breastfeeding and other feeding practices of UK mothers, as well as a set of other figures which may impact on their feeding choices (McAndrew, et al, 2012). The survey aimed to monitor current trends and inform policy. Beginning in 1975, it collected data every 5 years but ceased in 2015.
Sample
This study analyses the latest available figures, the 2010 series, which collected data on feeding practices at three stages post birth: 4-10 weeks (Stage 1), 4-6 months (Stage 2) and 8-10 months (Stage 3). An unclustered sample of 30 760 births was selected from all UK births registered in the period of August to October 2010. Data was weighted and corrected for both differential sample and differential response rates among the various groups (McAndrew et al, 2012).
The sample for this study was drawn from Stage 1, a total of 15 724 questionnaires that represented a response rate of 51%. SPSS v21 was used to transform and analyse the data and two filters were applied to even the sample and make results comparable and generalisable. The final sample resulted in 15 492 cases.
Measurement
For this secondary analysis, two dichotomy-dependent variables were formulated: bottle-feeding from birth and breastfeeding discontinuation. The bottle-feeding from birth variable referred to those mothers who always gave infant formula to their children, and never gave them breastmilk or put their babies to the breast. Breastfeeding discontinuation represented all mothers who attempted breastfeeding at some point but at the time of the survey were solely bottle-feeding.
The chosen independent variables were divided into three categories: socioeconomic and demographic factors (age, marital status, age of leaving full time education, socioeconomic classification, and ethnicity); family-related factors (parity, how their friends fed their babies); and birth factors (mode of birth and health problems after birth).
Data analysis
Inferential statistics were used to explore the association between the variables in the dataset, drawing inferences and unveiling the relationships between variables. Binary logistic regression analysis was used to clarify most of the relationships observed in the exploratory analysis, examining the effect of the above variables and representing the dependence structure.
The proportions (odds ratios) for both bottle-feeding from birth and breastfeeding cessation are presented in three sequential logistic regression models (Table 1). These models examine the four research questions addressed in the paper. Model 1 analysed predictors of both bottle feeding from birth and breastfeeding cessation, adjusting for demographic and family related variables (marital status, age, and parity). Ethnicity, socioeconomic classification, and age of leaving full-time education were added in model 2. Model 3 included birth-related factors (mode of birth, health problems after birth) and peer influence. P values (two-tailed) are presented for each result and only considered significant if P<0.05. Nagelkerke's pseudo R2 is presented to give an indication of the variance explained by each model. Figures are rounded to two decimal places.
| Bottle-feeding at birth (n=15 174) | Early breastfeeding cessation (n=11 795) | |||||
|---|---|---|---|---|---|---|
| Variable | Model 1 | Model 2 | Model 3 | Model 1 | Model 2 | Model 3 |
| Marital status | ||||||
| Married or in civil partnership (reference) | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Living together | 1.71† | 1.38† | 1.31† | 1.62† | 1.36† | 1.35† |
| Divorced, separated or widowed | 1.99† | 1.67∆ | 1.82∆ | 1.39 | 1.38 | 1.48 |
| Single | 3.38† | 2.32† | 2.24† | 2.08† | 1.84† | 1.83† |
| Age (years) | ||||||
| <20 | 5.54† | 2.69† | 2.36† | 3.59† | 2.45† | 2.45† |
| 20–24 | 3.26† | 2.16† | 1.96† | 2.39† | 1.89† | 1.99† |
| 25–29 | 1.79† | 1.49† | 1.39† | 1.60† | 1.50† | 1.55† |
| 30–34 | 1.25∆ | 1.20∆ | 1.14 | 1.09 | 1.09 | 1.09 |
| ≥35 (reference) | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Number of children | ||||||
| 1 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| 2 | 2.16† | 1.89† | 1.93† | 1.21 | 1.62† | 1.49∆ |
| 3 | 3.18† | 2.40† | 2.38† | 1.18 | 1.49∆ | 1.48∆ |
| 4+ | 3.90† | 2.40† | 2.39† | 1.18 | 1.39∆ | 1.39* |
| Ethnicity | ||||||
| White British | 6.10† | 4.80† | 2.30† | 2.56† | ||
| Northern Irish, Scottish | 12.11† | 8.65† | 3.69† | 3.92† | ||
| White any other | 1.72* | 1.57 | 0.99 | 1.23 | ||
| Mixed | 2.30∆ | 2.03* | 0.90 | 1.00 | ||
| Asian (reference) | 1.00 | 1.00 | 1.00 | 1.00 | ||
| Black | 0.66 | 0.64 | 0.19† | 0.20† | ||
| Socioeconomic classification | ||||||
| Managerial and professional (reference) | 1.00 | 1.00 | 1.00 | 1.00 | ||
| Intermediate occupations | 1.58† | 1.53† | 1.35† | 1.35† | ||
| Routine and manual occupations | 1.59† | 1.51† | 1.38† | 1.35† | ||
| Never worked | 2.62† | 2.57† | 1.26* | 1.28* | ||
| Not classified | 1.60† | 1.62† | 0.99 | 1.03 | ||
| Age of leaving full time education (years) | ||||||
| ≤16 | 2.62† | 2.39† | 2.10† | 2.03† | ||
| 17 | 2.17† | 2.04† | 1.80† | 1.74† | ||
| 18 | 1.78† | 1.64† | 1.60† | 1.54† | ||
| ≥19 (reference) | 1.00 | 1.00 | 1.00 | 1.00 | ||
| How did your friends feed their babies? | ||||||
| Most breastfed only (reference) | 1.00 | 1.00 | ||||
| Most gave infant formula only | 3.73† | 2.38† | ||||
| Most mix-fed | 1.74† | 2.41† | ||||
| A real mixture of the above methods | 2.92† | 2.32† | ||||
| No friends with babies | 2.35† | 2.21† | ||||
| Mode of birth | ||||||
| Vaginal birth (reference) | 1.00 | 1.00 | ||||
| Ventouse birth | 1.21 | 1.10 | ||||
| Forceps birth | 1.24* | 1.13 | ||||
| Caesarean section | 1.37† | 1.38† | ||||
| Health problems after birth | ||||||
| No (reference) | 1.00 | 1.00 | ||||
| Yes | 0.78∆ | 2.16† | ||||
| Omnibus χ2 | 1571.09 | 2629.16† | 2849.68† | 676.34† | 1448.79† | 1773.16† |
| Nagelkerke R2 | 0.15 | 0.25 | 0.28 | 0.08 | 0.16 | 0.20 |
P<0.05* P<0.01∆ P<0.001†
The highest odds of bottle-feeding were observed among women aged 20 years or less (OR=5.54) when compared to those over 35; single women (OR=3.34) when compared to married women; from low socioeconomic background (never worked) (OR=2.62) when compared to women in managerial and professional occupations and poorly educated women (who left education aged 16 or under) (OR=2.62) when compared to those who left education aged 19 or over. In addition, high odds were observed among white British women (OR=6) when compared to Asian women; those with increased parity of four or more (OR=3.9) when compared to first time mothers; women whose friends bottle-fed (OR=3.73) when compared to women whose friends mainly breastfed; and women who had caesarean sections (OR=1.37) when compared to women who had vaginal births.
Predictors of breastfeeding cessation were the same, except for socioeconomic classification, development of health problems post birth and parity. The highest odds of breastfeeding cessation were found among women from intermediate (OR=1.35) and from routine and manual (OR=1.38) occupations when compared to women in managerial and professional occupations and among women who developed health problems post birth (OR=2.16) when compared to those who did not. Parity was not found to be a significant variable in breastfeeding cessation.
A significant drop in odds ratio in the variable age in model 2 suggests that the added variables (ethnicity, social background, and education) mediated the relationship between maternal age and the dependent variables. Also of note are the differences in marital status, in which women who classified their status as ‘living together’ presented significantly higher odds of both bottle-feeding from birth (OR=1.71) and of early breastfeeding cessation (OR=1.62) than married women. In ethnicity, there is a decrease in the odds when adjusting for peer influence and birth-related factors (model 2 to model 3) in all ethnic categories in the bottle-feeding from birth group. Interestingly the opposite happens for the prediction of breastfeeding cessation.
Women whose friends were exclusively bottle-feeding had significantly higher chances of bottle-feeding from birth themselves (OR=3.34). In addition, mothers whose friends had mostly mix-fed were almost 2.4 times more likely to discontinue breastfeeding than those whose friends breastfed. The effect of mode of birth on feeding choices is challenging to interpret and results were only statistically significant for mothers who had a caesarean section.
The overall predictive power of the models in both analyses can be assessed by comparing the change in the Nagelkerke's pseudo R2, which has a greater increase between model 1 and 2 than model 2 and 3 (Table 1). This means that the variables included in model 2 have a greater contribution to explaining the variance in the model than the variables added in model 3.
Discussion
The majority of the results related to socioeconomic and demographic factors were in agreement with previous research. Being younger, single, poorly educated and unemployed were found to be strong predictors of bottle-feeding. This is consistent with research on breastfeeding, which reported that older maternal age, being married, higher education levels and professional occupations were strong breastfeeding predictors (Barnes et al, 1997; Shepherd et al, 2000; Al-Sahab et al, 2010; McAndrew et al, 2012).
Age
Young women are still developing their identity, negotiating and struggling with social interactions, trying to ‘fit in’ to society (Erikson and Erikson, 1997), and possibly reluctant to seek information and help. In addition, this group of women is also less informed (as their education is likely to still be in development); likely to still be single (and consequently, less well supported) and generally less prepared for pregnancy and motherhood (Reder, 2003).
Relationship status
Single motherhood in all ages was identified in the Marmot report (Marmot et al, 2010) as a cause of disadvantage in health, and was associated with lack of support and financial stability. This study confirmed that single women had a significantly higher chance of bottle-feeding from birth than married women. This was also the case for breastfeeding cessation, although with a lower impact.
Another noteworthy finding concerns living status. After adjusting for all other variables in the model, it was found that women living together with their partners were more than 30% more likely to report bottle-feeding from birth or early breastfeeding cessation than married women. This raises questions about the differences between the couples who cohabit and those who are married, and how this influences bottle feeding. Goodman and Greaves (2010) suggested that many of these differences are due to the stability of the relationship but they added that pre-existing characteristics, such as age, education, and other aspects of the family structure (such as whether the pregnancy was planned) predispose that stability. Less stability might be linked with a lack of support among the couples who cohabit.
Ethnicity
In a study by McAndrew, et al (2012), Northern Irish, Scottish and white British women were identified as the least likely to breastfeed, while research by Kelly et al (2006) found that Black and Asian women were the least likely to bottle-feed. This study confirms these findings and adds that a similar pattern with a lower magnitude is observed in breastfeeding cessation. Moreover, this study revealed that the results for all ethnic categories for are mediated by socioeconomic status for breastfeeding cessation—but not for bottle-feeding from birth. This suggests therefore that bottle-feeding could be more of a cultural choice than an economic one.
Furthermore, breastfeeding may not be perceived the same way among different cultures. Health beliefs are socially constructed and therefore a consequence of history, social interactions, policy and circumstance (Green and Thorogood, 2009). A possible explanation for such high odds of bottle feeding among white British women is a strong bottle feeding culture in the UK (Renfrew, et al, 2012) and a lack of governmental commitment to promote and protect breastfeeding (WBTi Steering and Core Groups, 2016). Previous studies have identified factors such as the support of family female figures, being breastfed or bottle-fed oneself, and witnessing positive or negative breastfeeding experiences as determinant influences in women's feeding intention (Hoddinott and Pill, 1999; Hoddinott, et al, 2010; Wilkins, et al, 2012). Many of the mothers and grandmothers of today's childbearing British women are likely to have fed their children with formula, which makes this method the most familiar.
Influence of friends
It is then also unsurprising that the multivariate analysis identified a strong effect of friends' bottle-feeding practices on bottle-feeding from birth, and friends' mixed feeding practices on breastfeeding cessation. These findings make sense in both a behavioural and a biological perspective. As discussed above, women are highly influenced by their family and peers, but biologically, mix-feeding is a very difficult to sustain as breastmilk is produced on a supply and demand basis (Marshall and Raynor, 2014). If a baby is given formula in some feeds, the breast is not stimulated, and the milk supply will reduce greatly or even cease.
Employment status
Being unemployed was robustly associated with bottle-feeding from birth, although it was women from intermediate and from routine/manual occupations who had the highest odds of breastfeeding cessation. These findings support the argument that people in lower socioeconomic positions adopt less healthy behaviours (Pampel et al, 2010): in this case, that they opt to bottle-feed from birth. But, as previously identified, the context in which they make this decision is likely to be in combination with other factors that affect their level of health consciousness and their health beliefs (Wardle and Steptoe, 2003). Bottle-feeding is endemic in some low-income communities where breastfeeding is rare, therefore:
‘Not being breastfed is both a consequence and a cause of social inequalities’
The combination of all these factors places this group of mothers at a disadvantage, resulting in an unequal chance of health for themselves and their babies.
Previous experience
Claims that failed experiences at breastfeeding lead mothers to opt for bottle feeding from birth in subsequent pregnancies (Jones, 1987; Taylor, et al, 2010) are confirmed by the findings of this study. A linear gradient was observed with the increase in the number of children, indicating that previous experiences with children led women to bypass attempts to breastfeed and to prefer bottle-feeding from birth. In their study of mothers' experiences after breastfeeding cessation, Larsen and Kronborg (2013) discussed feelings of guilt and failure that mothers experienced when they were unsuccessful and how that led to low self-efficacy and a fear of failing again. However, it is also important to note that not all bottle-feeding from birth cases relate to previously failed experiences. Previous research (Taylor, et al, 2010) has shown that women tend to repeat positive feeding experiences that they had with their first child with subsequent children. If they succeed on bottle-feeding in a first experience, they are likely to repeat it. In contrast, an increase in the number of children diminishes the odds of early breastfeeding cessation. This suggests that past experiences may also equip mothers with the ability to breastfeed their children for longer.
Mode of birth
The effect of mode of birth on feeding choices is challenging to interpret, as results were only statistically significant for mothers who had had a caesarean section. The results identified having a caesarean section as a predictor for both bottle-feeding from birth and breastfeeding cessation. These findings mirror the work of Ever-Hadani et al (1994) and Ahluwalia et al (2012), who found that having a caesarean section had a negative impact on breastfeeding. They theorised that it could be associated with the presence of health problems after birth, as caesarean sections have a greater potential to develop health complications than vaginal births (Marshall and Raynor, 2014). In this study, we can theorise the same for the breastfeeding cessation cases, where the presence of health problems after birth was a contribution factor for early breastfeeding cessation; however not for the bottle-feeding from birth mothers. This once again suggests that this decision is deep rooted.
To finalise, Nagelkerke's pseudo R2 has a greater increase between model 1 and 2 than models 2 and 3. This meant that the variables included in model 2 (ethnicity, socioeconomic status, and education) had a greater contribution to explaining the variance in the model than the variables added in model 3.
Conclusions
This study found strong independent associations between multiple socioeconomic, demographic, birth-related and peer influence factors for both bottle-feeding from birth and breastfeeding cessation. Bottle-feeding from birth was predicted by a range of socioeconomic and demographic disadvantages—namely being young, single, unemployed, white British, and being poorly educated, but also by other characteristics such as having more than one child, having a caesarean section, not having any health problems that affect the ability of the mother to feed her baby the way she intended, and having friends who bottle-fed their babies.
Early breastfeeding cessation was predicted by exactly the same determinants, except for socioeconomic classification, where working in intermediate or routine/manual occupations are the socioeconomic determinants, and developing health problems that affect the ability of the mother to feed her baby the way she intended.
As identified, the context in which mothers live was a key determinant in how they fed their babies. But in the opinion of the author, for many women, the decision either to bottle-feed from birth or to discontinue breastfeeding is not an option. Being young, not having role models, not having positive breastfeeding experiences, not having a supportive partner, not having a good socioeconomic position or good education all indicate that these decisions are not necessarily voluntary.
Recommendations
It is unfortunate that there is not a concerted effort to tackle the high bottle-feeding rates in the UK. The Government's ‘one-size-fits-all’ approach for increasing the breastfeeding rates, based on the assumption that bottle-feeding is a choice, will surely not have the recommended outcome of 6 months exclusive breastfeeding for all babies.
Governing bodies could make attempts to normalise breastfeeding by making changes that affect the culture in general, across all classes and all ages. One solution could be to include breastfeeding in the national curriculum, alongside reproduction and sex education. The delineation of targets for hospitals, councils, primary care trusts and universities to complete the baby-friendly initiative stages also seems necessary. Changing signs that identify baby-feeding facilities, often depicted by a baby and a bottle, to a ‘breastfeeding welcome’ sign, is another example of promotion. The Infant Feeding Survey, or a new version, should be re-introduced to collect data essential to the creation of interventions to promote, support and protect breastfeeding. This is not to mention deep and meaningful governmental statements that are also needed, such as changes in the labour law on maternity pay up to 6 months, and breastfeeding breaks or paid time for mothers who breastfeed past 6 months. The more that breastfeeding becomes normal and familiar, the more likely women are to adopt it.
The women that discontinued breastfeeding before they intended identified that more support from family and breastfeeding experts could have prevented the breastfeeding problems that they encountered (McAndrew, et al, 2012). Further research is needed on this topic, with a view to empowering women to become agents of their own choices through adequate support.
More can be done to tackle the high bottle-feeding rates in the UK, as it appears to be predominantly a cultural issue. A politically transparent message and consorted action is essential. The results of this study should be used to inform policy makers in order to create a clear and objective national breastfeeding strategy that supports the needs of women and their families, and seeks to normalise the breastfeeding culture in the UK.