Abstract

Background

Identifying the behavioral determinants of breastfeeding is an important step toward increasing breastfeeding rates, but studies often adopt a limited measurement model. We aimed to identify, in a British population, which behavioral and psychological factors, assessed throughout the perinatal period, were most reliably associated with intent to breastfeed and breastfeeding at 1 and 6听months.

Methods

This is an observational longitudinal study of a diverse (35.1% non-white) community sample of N鈥�=鈥�222 pregnant women attending a maternity hospital in the UK. We assessed self-reported anxiety and depressive symptoms, stressful life events, and coping at 20鈥�22, 28 and 36听weeks gestation and 1 and 6听months postnatally; intention to breastfeed was assessed at 20听weeks gestation. Breastfeeding was assessed at one and six months post-partum. We modelled the associations with logistic regressions, adjusting for socio-demographics.

Results

Antenatal and post-partum depressive and anxiety symptoms were not reliably associated with breastfeeding behavior up to 6听months. In contrast, breastfeeding intention, which was not associated with affective symptoms, stress, and coping, was a reliable predictor of breastfeeding after adjusting for covariates. The association between intention to breastfeed and breastfeeding behavior was not moderated by behavioral/psychological factors (p鈥�&驳迟;鈥�0.5).

Conclusions

This study extends previous findings about the importance of intention to breastfeed to breastfeeding behaviour and suggests that suffering from affective symptoms does not inhibit breastfeeding. Antenatal intention to breastfeed can play a crucial role in shaping both maternal and child health outcomes.

Background

There is strong evidence that human milk is the best diet for infants, providing vital nutrients to support growth and development in early life; breastfeeding duration and maintenance have been associated with improved nutritional, immunological, and social outcomes [1]. Specifically, exclusive and continued breastfeeding may help prevent a number of health issues, including respiratory problems, ear and bacterial infections [2], diarrhoea [3], and allergies in infancy [4] and diabetes [5], heart problems [6] and obesity [5] in adulthood. There are also physical and mental health benefits to breastfeeding mothers [7]. For instance, breastfeeding-induced oxytocin release has been associated with decreased stress and cortisol levels in mothers [8]. Nonetheless, breastfeeding rates are lower than is recommended by public health agencies [9, 10], e.g., the WHO recommends exclusive breastfeeding until 6听months followed by the introduction of complementary foods, and with continued breastfeeding up to two years old or beyond. Among European countries, the UK has one of the lowest breastfeeding rates [11] with a high initiation (81%), but decreased breastfeeding continuation (55% at 6听weeks) [12]. Cultural (e.g., gender role ideology), socio-economic (e.g., educational attainment), socio-political (e.g., breastfeeding policies), occupational (e.g., employment type and flexibility) and other health-related factors (e.g., time in the neonatal intensive care unit) have been linked to variation in breastfeeding practices [13]. Alongside these factors are psychological and behavioral factors [7, 14] which, to the extent that they are more modifiable than socio-demographic factors, may hold particular relevance for interventions to increase breastfeeding. Capitalizing on a longitudinal UK pregnancy cohort, the current paper examined competing psychological and behavioral factors collected throughout the perinatal period as predictors of breastfeeding until 6听months post-partum.

Several different behavioral and psychological factors may influence breastfeeding [15, 16]. Antenatal and/or post-partum mood disorders are among the most cited psychological factors associated with reduced breastfeeding success [17], but findings are inconsistent [18,19,20], perhaps due to different sample characteristics, small sample sizes, analysis of non-clinical symptom severity, variation in the psychological measures used, and insufficient control for confounders [17]. There is also variation in the timing of affective symptoms, with some suggesting antenatal risk and others focusing instead on post-partum risk from affective symptoms [19, 21]. The are also questions about possible directions of effect. For example, recent meta-analyses found that antenatal depressive symptoms were associated with decreased rates of exclusive breastfeeding [22] while breastfeeding was associated with a 14% lower risk of developing post-partum depression [23].

Other psychological and behavioral factors have received less attention as predictors of breastfeeding. Exposure to stressful life events (SLEs) and coping skills may be associated with breastfeeding, either directly or indirectly [24, 25]. For example, SLEs experienced听in the first gestational听trimester were associated with delayed onset of partial breastfeeding [26], whereas in others, financial [27] and traumatic stress [24] predicted early termination of breastfeeding. An additional study reported that antenatal stressful events were associated with reduced exclusive breastfeeding at three months [25]. However, the assessment of stress alone may be insufficient since the manner in which pregnant women cope with SLEs may determine outcomes like mental health [28] and breastfeeding decisions. Accordingly, measuring coping skills may clarify the impact of stress and symptoms on breastfeeding behavior. Coping has received limited attention in studies of perinatal health and behaviour [29], but inadequate coping skills including avoidance- or anger-based strategies have been associated with negative outcomes for mothers and infants [30, 31]. Here we test the novel hypothesis that antenatal coping skills are associated with breastfeeding practices up to six months, directly or indirectly, by modifying the impact of stress or symptoms.

In contrast to the somewhat inconsistent findings for affective symptoms and stress and coping, there is reliable evidence that intention to breastfeed robustly predicts later breastfeeding behavior [32]. Intention is a psychological (cognitive) dimension whose link with breastfeeding derives from a different conceptual basis. For example, the 鈥淭heory of Reasoned Action鈥� (TRA) [33] suggests that a primary determinant of behavior is the individual intention to perform it. This model has attracted interest from some studies of breastfeeding outcomes [14]. Most women (82鈥�97%) decide on the infant feeding method during pregnancy [34, 35], with the majority of those who intend to breastfeed carrying this out [36]. An alternative focus for observational and intervention studies of breastfeeding could be breastfeeding intention, given that it is a comparatively simple measure. For example, analysis of a large community cohort study of over 9,000 participants found that a positive antenatal attitude towards breastfeeding was associated with a 20鈥�30% increase in breastfeeding initiation and maintenance [32]. However, intention to breastfeed may be confounded with, and a proxy for, perinatal psychological distress which shapes breastfeeding behavior, such as antenatal anxiety [37]. Alternatively, the impact of intention on breastfeeding practices may be moderated by affective symptoms. We test these novel and competing hypotheses in the current study.

In summary, we aimed to examine behavioural and psychological factors, assessed throughout the perinatal period, as predictors of breastfeeding at 1 and 6听months; we also test exploratory moderation hypotheses derived from the stress and coping literature [38] and not previously examined in relation to intention to breastfeed and breastfeeding outcomes. The current study adds to the literature by assessing multiple competing psychological and behavioral measures as predictors of breastfeeding at 1 and 6听months and by testing for possible moderating effects of the breastfeeding intention and behavior relationship.

Methods

Study design

This is a UK observational prospective longitudinal study including pregnant women and following them until 6听months post-partum.

Participants and procedure

Participants were recruited at the antenatal clinic at Queen Charlotte鈥檚 and Chelsea Maternity Hospital (QCCH) in London, UK. Women listed by the antenatal clinic to attend their 20听weeks ultrasound control were approached by a doctoral student who provided oral and written information about the study. Women were eligible if they were between 20鈥�22听weeks gestation, aged between 18 and 45听years old, self-reported computer literate, fluent in English, and had access to a device with internet access. Exclusion criteria included severe medical disorders (e.g., epilepsy), a currently diagnosed severe psychiatric disorder (e.g., psychosis or drug addiction), high-risk pregnancy (e.g., preeclampsia), medically assisted reproduction, or multiple pregnancy. Postnatally, women who had a premature birth (鈮も��36听weeks) and/or a baby with a severe disorder (e.g., heart defect) were excluded. Physical copies of participants鈥� medical records were checked weekly by the student to confirm post-partum inclusion criteria as well as to collect birth and baby鈥檚 information.

Women who agreed to participate and provided written听informed consent听during recruitment (N鈥�=鈥�369), provided basic personal information (name, email address, home address, telephone, date of birth, baby due date, gestational age) and completed online surveys from pregnancy (T1: 20鈥�22 gestational weeks, T2: 28 gestational weeks, T3: 36 gestational weeks) to post-partum (T4: 1听month, T5: 6听months). At each subsequent time point, participants were sent an email containing a link to complete the online questionnaires. For those who did not complete the questionnaires, up to four email reminders were sent, one per week. The emails were sent by research assistants at each time point. The participants have received no compensation to take part in the study.

Sample size

From the 369 enrolled participants, 222 had available data on type of infant feeding at 1听month (see Fig.听1; see attrition analyses in results section). A post hoc power analysis was conducted with G* Power version 3.1.9.7. It indicated that with an N鈥�=鈥�222, we could achieve 99% power for detecting a medium effect (f鈥�=鈥�0.25), with significance of 伪鈥�=鈥�0.05, for all tests conducted.

Fig.听1

Flowchart diagram of participants withdrawal

Instruments and measures

Maternal depressive symptoms were assessed using the Edinburgh Postnatal Depression Scale (EPDS) [39], a 10-item self-report questionnaire designed to measure post-partum depressive symptoms. This measure, which has been successfully validated for assessments in the perinatal period, quantifies the intensity of symptoms within the last seven days. Items are rated in a 4-point Likert scale (0鈥�3), with a total score ranging from 0 to 30. We analysed EPDS as a continuous variable; higher scores indicate higher depressive levels. Depressive symptoms were assessed at 20鈥�22, 28 and 36听weeks gestation and at 1 and 6听months. The instrument shows 86% sensitivity, 78% specificity and 73% positive predictive value for a similar population [39, 40]. Cronbach鈥檚 alpha for this sample was 伪鈥�=鈥�0.86. The complete questionnaire was utilized for this study and, additionally we created a composite variable (Cronbach 伪鈥�=鈥�0.91) by averaging the results from the three time points of antenatal assessments.

Anxiety symptoms were measured with the State and Trait Anxiety Inventory (STAI) [41]. The STAI comprises two separate scales 鈥� State (S) and Trait (T) 鈥� each with 20 items measured in a 4-point Likert scale. The State scale asks the participants to rate how they feel at the specific assessment time in terms of intensity (鈥渘ot at all鈥� to 鈥渧ery much so鈥�) whereas the Trait scale asks participants how they generally feel in terms of frequency (鈥渁lmost never鈥� to 鈥渁lmost always鈥�). Sample items include 鈥淚 feel upset (STAI-S)鈥�, 鈥淚 am a steady person (STAI-T)鈥�. We assessed it as a continuous variable in which higher scores indicate higher anxiety. Anxiety symptoms were assessed at 20鈥�22, 28 and 36听weeks gestation and at 1 and 6听months. The instrument yielded, on both state and trait scales, optimal sensitivity (80%), specificity (79.75%), and positive predictive value (51%) for a similar population. The complete questionnaire was utilized for this study and, additionally we created a composite variable (Cronbach 伪鈥�=鈥�0.98) by averaging the results from the three time points of antenatal assessments.

The Stressful Life Events (SLEs) scale used was created based on the Holmes-Rahe stress inventory [42] assessing 24 SLEs. SLEs in pregnancy were based on responses since they got pregnant (20鈥�22, 28 and 36听weeks) and in the post-partum period (1 and 6听months). Participants were asked if the event occurred and, if so, how severely affected they were by it (0鈥�=鈥�No, this never happened, 1鈥�=鈥�Yes, this has happened and it affected me a little, 2鈥�=鈥�Yes, this has happened and it affected me a lot). Sample items include 鈥淵ou had a major financial problem鈥�. We assessed it as a continuous variable in which higher levels indicate increased stress. Cronbach鈥檚 alpha for this sample was 伪鈥�=鈥�0.60. The complete instrument was utilized for this study and, additionally we created a composite variable (Cronbach 伪鈥�=鈥�0.85) by averaging the results from the three time points of antenatal assessments.

Coping styles were measured with the English version of the Utrecht Coping List-19 (UCL-19) [43, 44]. This 19-items questionnaire describes an individual tendency for five coping styles: emotional coping, avoidant coping, palliative coping, problem-focused coping and social coping. The 4-point Likert scale ranges from 0 (鈥渘ever鈥�) to 3 (鈥渧ery often鈥�) with high scores indicating a tendency to apply the specified coping style when facing unpleasant events. This questionnaire was successfully used in research involving pregnant and post-partum women and was filled out by our participants at all time points. Principal component analysis, supported by consideration of previous literature, and follow-up analysis that excluded items that loaded on multiple factors suggested four factors, which were labelled problem-focused coping, avoidant coping, emotion-focused coping and non-social coping. Internal consistencies, using Cronbach鈥檚 alpha, were acceptable for each factor: 伪鈥�=鈥�0.80 for problem-focused coping (e.g., 鈥淒o you work goal-directed to solve a problem?鈥�), 伪鈥�=鈥�0.79 for non-social coping (e.g., 鈥淒o you seek comfort and sympathy?鈥�), 伪鈥�=鈥�0.70 for emotion-focused coping (e.g., 鈥淒o you show your frustration?鈥�) and 伪鈥�=鈥�0.54 for avoidance coping (e.g., 鈥淒o you avoid difficult situations as much as possible?鈥�). Cronbach鈥檚 伪 was similar across all time points. All four sub-scales (problem-focused coping, avoidant coping, emotion-focused coping and non-social coping) at each time point were used in this study and, additionally we created a composite variable (Cronbach 伪鈥�=鈥�0.90) by averaging the results from each sub-scale from the three time points of antenatal assessments.

Intention to breastfeed was assessed at 20鈥�22听weeks gestation with the question 鈥淲hat type of infant feeding are you intending to provide to your baby?鈥�. Potential answers included breastfeeding, formula feeding, mixed feeding and unsure. For this study, responses were scored as intention to breastfeeding (yes vs. no/unsure).

Breastfeeding was assessed at one and six months post-partum with the question: 鈥淗ow are you currently feeding your baby?鈥�. Response choices were human milk only, combined human milk and formula feeding, and formula feeding only. We define breastfeeding at 1听month as exclusive (i.e., human milk only) and partial (i.e., combined human and formula feeding) breastfeeding. Breastfeeding persistence was defined as exclusive breastfeeding (i.e., human milk only) at 6听months; we also ran supplementary analyses examining partial breastfeeding (i.e., combined human and formula feeding) at 6听months.

Socio-demographic information was collected at 20鈥�22听weeks gestation and birth and baby characteristics were obtained from participants鈥� medical records. Key variables were: marital status (married/cohabitating vs. single), ethnic/race group (although overall ethnically/racially diverse鈥�6.4% Asian or Asian British, 3.6% Black, Black British, Caribbean or African, 5.3% Mixed or multiple ethnic groups, 19.8% Other ethnic groups鈥攖he small number of most race and ethnic categories necessitated coding this as white vs. non-white), gravidity (number of times that a woman has been pregnant, discrete variable), parity (number of times that she has given birth to a fetus with a gestational age of 24听weeks or more, continuous variable), monthly income (ranging from鈥夆墹鈥�18,000 拢 to鈥�>鈥�100,000 拢), self-reported current smoking (any smoking, yes vs. no), self-reported current alcohol consumption (any level of consumption, yes vs. no), maternal education (university degree or higher vs. no university degree), currently working (yes vs. no), delivery mode (caesarean section vs. other), maternal age at delivery (continuous variable), child sex (boy vs. girl) and neonatal weight (continuous variable). We included covariates which were previously found to be robust predictors of breastfeeding on an a priori basis, namely parity, ethnicity [45] education [46] and mode of delivery [47].

Our outcomes consist of the intention to breastfeed and breastfeeding behaviour measured through multiple choice questions, whereas our exposures were depressive symptoms, anxiety symptoms, SLEs and coping measured through the above-mentioned questionnaires. Our analyses also controlled for sociodemographic and birth and baby characteristics obtained through a questionnaire and medical records.

Ethics considerations

Ethical approval was obtained from the Imperial College Healthcare Trust ethics committee and NHS Health Research Authority ethics committee before commencement of the study (Approval No. 13EE0059). This study and all its experiments were conducted in accordance with the principles of the Declaration of Helsinki.

Statistical analyses

Descriptive statistics including frequency, percentage, median and interquartile range were used to summarize the study data. Since most of the affective symptoms and coping sub-scales were skewed, we applied log transformation using the formula log10 (value鈥�+鈥�1). Visual inspections and Kolmogorov鈥揝mirnov tests showed that some variables were still not normally distributed, but the skewness of the data was improved; therefore, non-parametric tests (e.g., Spearman correlations) were also conducted. We created composite variables for anxiety symptoms, depressive symptoms, stressful life events and maternal coping for the antenatal period by averaging the results from the three time points of antenatal assessments. However, post-partum measures, namely depressive symptoms, were not combined so that it was possible to assess concurrent and longitudinal associations between psychological measures and breastfeeding behavior. Spearman correlations were conducted between covariates, affective symptoms, maternal coping, intention to breastfeed and breastfeeding at one and six months. The aforementioned tests were used to analyse our two research aims.

Binomial logistic regression was our primary analytic strategy to analyse our first aim; analyses of exclusive and partial breastfeeding were done at one-month; following WHO recommendations, we defined persistence as exclusive breastfeeding through six months. All potential predictors of breastfeeding at 1 and 6听months were entered in one single model together with the covariates. Given our interest in assessing intention to breastfeed independently from other demographic and psychological/behavioral variables, we included education, ethnicity, antenatal and post-partum depression, and antenatally assessed coping and stressful life events on an a priori basis; other covariates were included if they were associated with breastfeeding outcomes in bivariate analyses. The final regression model consisted of education, ethnicity, antenatal and post-partum depression, and antenatally assessed coping (problem-oriented) and stressful life events.

To explore the impact of behavioural and psychological factors on the breastfeeding intention and breastfeeding behaviour relationship, we performed a moderating effect analysis. For this, we created interaction terms between breastfeeding and behavioural and psychological variables (antenatal depressive symptoms, antenatal anxiety symptoms, stressful life events, coping) and added them in the logistic regression models after including main effects. The final model included education, ethnicity, antenatal and post-partum depression, and antenatally assessed coping and stressful life events and one of the above-mentioned interaction terms (breastfeeding intention * antenatal depressive symptoms). A two-sided P-value of鈥�<鈥�0.05 was used to assess statistical significance. Statistical analyses were performed using the IBM Statistical Package for the Social Sciences (IBM SPSS Statistics for Windows Version 24.0).

Results

Data was collected from April 2012 to April 2013. Sample characteristics are shown in Table听1. There was modest evidence of selective attrition from enrolment until 6听months. We tested for selective attrition by creating a variable indicating the number of completed visits听 (scores range from 1 to 5) and using this group variable to comparing scores on the psychological measures at enrolment; analyses were based on ANOVA. Of the eight psychological measures collected at enrolment, two (EPDS, STAI) were associated with retention, but in neither case was there evidence of a linear pattern (e.g., the number of completed visits was not linearly related to initial symptoms): analyses for EPDS a significant omnibus test (F(4,368)鈥�=鈥�2.85, p鈥�=鈥�0.02) but none of the post hoc pairwise contrasts was significant at p鈥�&濒迟;鈥�0.05. Similarly, for STAI, there was a significant omnibus test (F(4,368)鈥�=鈥�2.64, p鈥�=鈥�0.03) but none of the post hoc pairwise contrasts was significant at p鈥�&濒迟;鈥�0.05.

Table听1 Sociodemographic characteristics of the sample (N鈥�=鈥�222)

Most women were married or cohabiting (73.4%), had a university degree or higher (84.2%), had higher than UK general population mean income (拢23鈥�600 in 2013); approximately one-third were non-white. More than half of the sample had a natural birth. There were no significant differences between women who supplied information on type of infant feeding given at 1 (n鈥�=鈥�222) and 6听months (n鈥�=鈥�164). Antenatal (Median: 6.0, IQR: 5.9) and post-partum (Median: 6.0, IQR: 5.7) depressive symptoms showed similar values both for all women and for those who exclusively breastfed. Similar values were also found for anxiety state (Median: 31.1, IQR: 14) and trait (Median: 32.6, IQR: 12.3). In general, women described having less than one SLEs (Median: 0.6, IQR: 2).听Table 2 reports the breastfeeding characteristics of the sample.

Table听2 Breastfeeding characteristics of the sample

Bivariate associations are reported in Table听3听and indicate that antenatal intention to breastfeed was associated with exclusive breastfeeding (r_s鈥�=鈥�0.20, P鈥�<鈥�0.05) and partial breastfeeding at 1-month (r_s鈥�=鈥�0.25, P鈥�<鈥�0.05). Breastfeeding intention was not, however, significantly correlated with any of the symptoms, stress, or coping measures (all correlations鈥�<鈥�0.10), but was higher in the white group. For neither the antenatal nor post-partum period was there a significant association between breastfeeding and depression, anxiety, stressful life events or coping. We nonetheless included antenatal and 1听month post-partum depression in the primary prediction model because of our interest in testing the specific, independent effects of intention on breastfeeding behavior.

Table听3 Correlation matrix between predictors and outcomes adjusted for covariates

A binomial logistic regression indicates that women who intended to breastfeed antenatally were 4.5 times more likely to initiate breastfeeding and provide partial breastfeeding through 1听month (95% CI 1.63鈥�12.95, P鈥�<鈥�0.01, see Additional file 1) 鈥� after adjusting for covariates, including antenatal and 1听month post-partum depression and socio-demographic factors. The prediction from intention to breastfeeding was weaker for exclusive breastfeeding persistence until 6听months (Table听4): a marginally significant association was found between intention to breastfeed and exclusive breastfeeding at 6听months (95% CI 0.88鈥�19.42, P鈥�<鈥�0.07, OR 4.13). Similar results were obtained when alternative psychological measures (coping, anxiety, stressful life events) were included in the model as covariates.

Table听4 Logistic regression analyses between predictors and exclusive breastfeeding outcomes at 1 and 6听months

Additional analyses indicated no significant evidence (p鈥�>鈥�0.05) that the prediction from intention was moderated by maternal antenatal or post-partum affective symptoms or stress. Supplementary analysis also indicated no moderation of intention on breastfeeding at 1- or 6-months associated with race (white compared with non-white) or education (university degree compared with no university degree).

Discussion

We assessed a wide range of psychological and behavioral factors in the antenatal and post-partum period that had been suggested to be associated with breastfeeding, including maternal affective symptomatology, coping strategies, infant feeding intention and SLEs as potential correlates of breastfeeding behaviour in a UK community sample. Three main findings emerged: 1) antenatal intention to breastfeed was a strong predictor of breastfeeding at one month; the effect was less evident by six months; 2) antenatal and post-partum depressive and anxiety symptoms, stressful life events, and coping styles were not reliably associated with breastfeeding intention or breastfeeding behaviour until 6听months; 3) the association between intention to breastfeed and breastfeeding outcomes was not moderated by symptoms and coping factors.

Our results extend previous studies by suggesting that intention to breastfeed, which is easily assessed in the antenatal period, can be a robust predictor of breastfeeding at 1听month [48, 49] 鈥� independent of affective symptoms, stress, and coping. A key advantage of this study is the repeated measurement of psychological and behavioral dimensions throughout the pre- and postpartum period. That provided substantial coverage of the psychological and behavioral determinants of breastfeeding that were assessed in prior studies. Additionally, a novel feature of the study is that we included measures of coping styles, which needs to be considered in studies of parenting [50]. However, we did not find reliable associations between coping, assessed in the antenatal period, and breastfeeding intention or breastfeeding behaviour. Similarly, we did not find evidence that affective symptoms, coping or stress moderated the associations between breastfeeding intention and breastfeeding behavior.

The lack of association between affective symptoms and breastfeeding does contrast with some prior studies [22, 23]. On the other hand, findings on antenatal depression are inconsistent, with some studies showing that it predicted early cessation of breastfeeding [51] or reduced breastfeeding initiation, but others failing to replicate it [35]. A large-scale UK community cohort of more than N鈥�=鈥�9000 mothers found no significant difference in breastfeeding initiation and duration between those who reported depressive symptoms in the听antenatal period听[32]. It may be that inconsistencies across studies are explained by sample composition and attention to potential confounders, such as intention, perinatal risk, family context, and occupational status. The somewhat inconsistent findings concerning affective symptoms and stress with breastfeeding contrast with the consistent finding that intention is strongly associated with later breastfeeding [52]. A key and more novel observation in this study is that breastfeeding intention is distinguishable from 鈥� indeed essentially independent of 鈥� affective symptoms, stress, and coping.

Antenatal intention to breastfeed plays a crucial role in shaping both maternal and child health outcomes. Research demonstrates that a mother鈥檚 intention to breastfeed may be linked to her knowledge, self-efficacy and attitudes towards infant feeding [49, 53]. Moreover, the significance of antenatal breastfeeding intentions extends beyond the actual breastfeeding experience; these intentions might reflect a higher commitment to children鈥檚 overall health [54]. Mothers who intend to breastfeed may engage in healthy behaviors related to sleep, nutrition, exercise, and medical care. They are more likely to adopt additional health-promoting practices that benefit fetal development and can mitigate the risks associated with infections [55]. Even in cases where breastfeeding does not occur, these mothers often engage in compensating behaviors鈥攕uch as selecting specialized formulas or homemade baby foods鈥攖o ensure optimal health for their infants [55]. Notably, this positive health-related behaviour in mothers who intend to breastfeed may depend on their socio-economic, socio-political, and occupational and family setting. However, by recognizing that these intentions align with broader maternal health behaviors and knowledge, healthcare practices can be better tailored to promote not only breastfeeding but also holistic maternal and child wellbeing.

Educational interventions to emphasize breastfeeding as the best nutritional choice for the physical and emotional health of babies have been in place for many years, but with limited success [56]. Interventions targeting malleable factors like breastfeeding intentions may offer significant opportunities. For instance, peer counsellor programmes, home visits by professionals and support to women鈥檚 feeding choices were shown to be effective in promoting breastfeeding during pregnancy [57]. Furthermore, our findings imply that intervention efforts targeting intention can be encouraged despite, and are unlikely to be impeded by, co-occurring affective symptoms and stress. Our findings also suggest that the presence of affective symptoms should not discourage breastfeeding or interventions to improve breastfeeding.

There are limitations to this study. The sample was recruited from one London Hospital serving the National Health System and the community sample was diverse, but weighted toward high levels of education, professional status and income than the UK. The high breastfeeding rates may also reflect hospital practice, as midwives鈥� training programmes at this hospital actively encourage breastfeeding. Thus, the findings from this study gives insight into this particular UK population group and may not generalize to high-risk samples and settings. Our results should be replicated in low-income populations where often family stress and lack of food to support breastfeeding may adversely affect maternal mental health and their ability to sustain breastfeeding despite their best intentions. Moderation analyses could have been impaired by insufficient power given that a large sample size is needed to find a moderation effect. In addition, we had limited data on socio-economic context, most notably maternal working conditions. Our focus instead is on psychological and behavioral predictors in order to identify the most robust and potentially modifiable factors. Also, there was attrition across the perinatal period, but this was not very strongly associated with initial distress. Finally, the study assessed breastfeeding on only two occasions until 6听months, which is consistent with many previous studies; nonetheless, more intensive assessments and longer follow-up periods might yield different patterns of results. The study strengths include the longitudinal data collection involving multiple assessments, which enabled us to examine any exacerbation of affective symptoms, stress, and coping in relation to breastfeeding in a repeated fashion.

Conclusion

In a UK community sample, we found that antenatal intention to breastfeed was a robust predictor of breastfeeding, in contrast to a wide range of behavioral and psychological factors which were not. Neither was the association between intention to breastfeed and breastfeeding outcomes moderated by these behavioral and psychological factors. The implication is that suffering from perinatal affective symptoms is not a barrier to breastfeeding. Interventions targeting intention to breastfeed hold particular promise for increasing breastfeeding behavior and its positive impacts for the mother and child.