Association of breastfeeding with mental disorders in mother and child: a systematic review and meta-analysis – BMC Medicine

Synthesis

A total of 17,887 items were identified through initial searches, and 40 articles met the inclusion criteria after duplicate removal and title/abstract screening (Fig. 1). Full-text manuscripts were assessed, and 19 articles were excluded, resulting in 21 articles included in the qualitative synthesis, with 18 investigating association between breastfeeding and child health and three looking at maternal health outcomes. Of these, three studies were included in the quantitative synthesis for the outcome of schizophrenia. The included studies were published between 1997 and 2023 and were all observational, comprising of 6 case–control studies [27,28,29,30,31,32], 5 retrospective studies [33,34,35,36,37] and 9 prospective cohort studies [38,39,40,41,42,43,44,45,46] and one cross-sectional study with retrospective assessment of the exposure [47].

The studies investigated associations between breastfeeding and mental health disorders in children and maternal mental health disorders. Specifically, eight studies assessed depressive disorders [33, 34, 36, 40, 41, 43, 45, 47], six investigated schizophrenia [28,29,30, 32, 37, 38] and five looked at anxiety disorders [31, 41, 43, 45, 47], while eating disorders [42] and borderline personality disorder [27] were assessed in one study each. In one study, the prevalence of mental health disorders in children was described without specifying the outcome [44].

Three studies evaluated the associations between breastfeeding and maternal mental health disorders [35, 39, 46].

Participant characteristics

Studies were carried out in twelve different countries, with the majority of the research being conducted in the European and Australasian regions. There were three studies conducted in China [33, 34, 40], three in the USA [36, 39, 42], three in Australia [41, 43, 46], and in the United Kingdom [32, 37, 47], two in Brazil [44, 45] and one each in South Korea [35], Japan [29], South Africa[28], Italy [30], Turkey [31], Denmark [38] and Germany [27].

Sample size ranged from 160 [39] to 186,452 [25] participants in cohort studies and between 100 [28] and 450 [31] in case–control studies. Most of the studies followed children up to the adolescence with the maximum follow-up duration of 40 years.

Breastfeeding definition and reporting

There was a substantial variation in breastfeeding reporting and definitions used. Some studies collected already predefined data from the registries [46], others used structured standardised (e.g. Pre-/Postnatal Stress Questionnaire (NPQ-PSQ) [27], Growing Up Today Study (GUTS) questionnaire [42]) and non-standardised questionnaires and interviews [28, 30, 41, 44, 45, 47, 31,32,33,34, 36, 37, 39, 40] obtaining the data prospectively or retrospectively. Exclusive breastfeeding was usually defined as breastfeeding without intake of “any other food”.

Association between breastfeeding and mental health in children

Schizophrenia

Six studies investigated schizophrenia, with four using case–control design [28,29,30, 32], and two cohort studies, a prospective [38] and a retrospective [37] (Table 1). The outcome of interest was defined in accordance with Diagnostic and Statistical Manual of Mental Disorders (DSM)-III [32], DSM-IV [28,29,30], ICD 9 [37], International Classification of Diseases (ICD)-8 code 295 or ICD-10 code F20 [28]. In a single study, Sorensen and co-authors defined schizophrenia as bizarre delusions, delusions of control, abnormal affect, autism, hallucinations and disorganised thinking [38].

The study risk of bias determined by means of NOS was good in cohort studies, while case–control studies were generally of satisfactory risk of bias (range satisfactory to good). The common flaw was lack of adjustment for potentially significant confounders such as family history of schizophrenia, with four studies presenting crude data only [28,29,30, 37]. Exposure to breastfeeding was self-reported by parents, and for several case–control studies, the recall period exceeded 20 years [32, 37, 38].

All but one study [38] found no association between breastfeeding and schizophrenia, Sorensen et al. reported an association between breastfeeding for two weeks or less and increased risk of schizophrenia adjOR 1.73 (95% CI 1.13–2.67) upon adjustment for maternal schizophrenia, single mother status, sex and parental social status at 1 year of age.

Heterogeneity in methodology and lack of relevant outcome reporting did not allow for meta-analysis of the data from cohort studies. The pooled data from three case–control studies (n = 528) showed no significant association between breastfeeding and schizophrenia later in life OR 0.98 (95% CI 0.57–1.71) [28,29,30] (Fig. 2). In the sensitivity analysis, addition of another study [37] which used siblings as a control group did not change the results OR 0.89 (95% CI 0.58–1.38). The GRADE certainty of evidence was very low due to risk of bias and serious imprecision (Table 2).

Depressive disorders

Five studies (three retrospective [33, 34, 36], two prospective cohorts [31, 36] and one cross-sectional study with retrospective assessment of the exposure which derived data from the UK biobank cohort [47]) investigated depressive disorders (Table 3).

A wide range of instruments was used across the studies for outcome assessment: Mini International Neuropsychiatric Interview for Children and Adolescents (MINI-KID) 5.0 [33, 45], Child Behaviour Check List (CBCL) which is very consistent with DSM-V diagnostic categories [34]. Other scales included PHQ-9 (self-reported Patient Health Questionnaire-9) that is consistent with DSM-IV [40, 47] and Schedule for Affective Disorders and Schizophrenia for School-Age Children (K-SADS) which is compatible with DSM-III [36]. According to PHQ-9, depression was defined as having a total score of 11 or more mapping on to DSM-IV [40]. De Mola et al. used Beck’s Depression Inventory (BDI-II) validated for Brazilian population to assess severity of depressive symptoms [45].

The study risk of bias was satisfactory on average, ranging between 4 and 7 as per NOS.

Kwok et al. and De Mola et al. reported no association between breastfeeding and depressive disorders development [40, 45] upon adjustment for multiple potential confounders. Other two studies showed that absence of any breastfeeding was associated with an increased risk (adjOR 1.88 (95% CI 1.28–2.49 and adjOR 1.71 (95% CI 1.14–2.56) respectively) of depressive disorders later on in life (6–16 years of age) [33, 36]. Huang et al. reported a protective effect of breastfeeding only in children who were breastfed longer than 6 months compared to those who have never been breastfed adjOR 0.45 (95% CI 0.23–0.91), while shorter duration of breastfeeding was not associated with any protective effect adjOR 0.79 (95% CI 0.37–1.67) [34]. A very recent cross-sectional study from Liu et al., based on the data from the UK Biobank cohort suggested protective effect of breastfeeding in mothers smoking adjOR 0.86 (95% CI 0.82–0.91) and not smoking adjOR 0.82 (95% CI 0.79–0.84) during pregnancy [47]. The overall GRADE certainty of evidence was very low due to potential risk of bias, serious inconsistency and indirectness (Table 2).

There was a lack of studies reporting the number of participants in the exposed and non-exposed groups to conduct meta-analysis.

Anxiety disorders

Three studies (one case–control [31], one prospective cohort study [36] and one cross-sectional study with retrospective assessment of the exposure using the data from the UK biobank cohort [47]) investigated association between breastfeeding and anxiety disorders [22, 36] (Table 4).

For the primary outcome assessment, Orengul et al. investigated social anxiety disorder, unspecified anxiety disorders, generalised anxiety disorder, specific phobias, separation anxiety disorder and panic disorder using Schedule for affective disorders and schizophrenia for school-age children, present version (K-SADS-P), and the Revised Child Anxiety and Depression Scale (RCADS), total anxiety subscale [31], while De Mola et al. studied generalised anxiety disorder and social anxiety disorder using Mini International Neuropsychiatric Interview version 5.0 validated for Brazil [45].

The risk of bias of the studies varied from unsatisfactory to good.

De Mola and co-authors in their cohort study found no association between breastfeeding or its’ duration and anxiety disorders development [45], while a case–control study from Orengul et al. [31] found reduced risk in breastfed children when compared with those who have never been exposed to breast milk OR 0.17 (95% CI 0.05–0.60). Liu et al. found that breastfeeding in mothers smoking adjOR 0.87 (95% CI 0.79–0.87) and not smoking adjOR 0.83 (95% CI 0.79–0.87) during pregnancy being associated with less anxiety in a cross-sectional study based on the data from the UK Biobank cohort [47]. The overall GRADE certainty of evidence was very low due to potential risk of bias and serious inconsistency (Table 2).

There was a lack of studies reporting the number of participants in the exposed and non-exposed groups to conduct meta-analysis.

Depressive/anxiety disorders as a composite outcome

Two cohort studies assessed anxiety and depression as a composite outcome. Hayatbakhsh and co-authors used the Youth Self Report (YSR) of the Child Behaviour Check List (CBCL) which has the same format as the CBCL but with questions paraphrased in the first person [41]. Oddy et al. reported composite outcome “internalising complaints” that included withdrawn, somatic complaints, anxiety, and depression [43]. Hayatbakhsh et al. found breastfeeding for at least 4 months to be associated with lower scores of CBCL “anxiety/depression” domain at 14 years of age, while Oddy et al. reported breastfeeding for less than 6 months being associated with higher risk of internalising complaints OR 1.21 (95% CI 1.0–1.46).

There was a lack of studies reporting the number of participants in the exposed and non-exposed groups to conduct meta-analysis.

Other mental health disorders

Other conditions investigated in the reviewed manuscripts included borderline personality disorder (BPD) [27] and eating disorders [42] (Table 5).

In a case–control study, Schwarze et al. investigated BPD defined as a pervasive pattern of impulsivity, emotional instability, identity disturbance and dysfunctional interpersonal relationships and diagnosed according to DSM-IV criteria for BPD [27]. Authors reported increased odds of BPD in those who have never been breastfed adjOR 4.68 (95% CI 1.88–11.66).

In a cohort study, Iron-Segev et al. assessed a broad range of eating disorders, including bulimic behaviours like purging, binge eating and other self-reported eating disorders like anorexia nervosa and bulimia nervosa. Purging was defined as using laxatives or force vomiting to lose weight or keep from gaining weight more than one time a month. The patient was considered as binge eater if eating binge at least once a month and feeling out of control while doing so was reported [42]. The study did not find any associations between breastfeeding duration (< 4 months, 4–9 months, > 9 months) and any eating disorder.

There was a lack of studies reporting the number of participants in the exposed and non-exposed groups to conduct meta-analysis.

Association between breastfeeding and maternal mental health

Three studies (a retrospective [35] and two prospective cohorts [39, 46]) investigated associations between breastfeeding and long-term development of maternal mental health outcomes post lactation (Table 6). Among them, one looked at a variety of mental disorders, including schizophrenia, unipolar depression, bipolar affective disorder and anxiety disorders [46] and two assessed associations with maternal depression [35, 39].

For depression assessment, instruments used included Patient Health Questionnaire 9 (PHQ-9), with scores of 10 or higher indicating depression [35], Edinburgh Postnatal Depression Scale (EPDS), with a cut-off of 10 scores being considered a minor depression [39] and ICD-10 codes at admission for each diagnosis of interest [46].

The study risk of bias determined by means of NOS varied from satisfactory to good.

Xu et. al reported association between absence of breastfeeding at the time of hospital discharge and higher risk of schizophrenia adjOR 2.0 (95% CI 1.3–3.1) and bipolar affective disorder adjOR 1.9 (95% CI 1.1–3.5) 12 months after delivery but found no protective effect against anxiety disorders and unipolar depression [46].

Park et al. reported protective effect of prolonged (> 47 months) breastfeeding against maternal depression in the postmenopausal period adjOR 0.33 (95% CI 0.16–0.68) [35].

Hahn-Holbrook and co-authors suggested that women who breastfed more frequently at 3 months postpartum showed greater subsequent declines in depressive symptomatology over time compared to women who breastfed less frequently and lower absolute levels of depressive symptoms by 24 months since birth [39].

Heterogeneity in outcome assessment did not allow for meta-analysis.