Article Text

Download PDFPDF

Maternal postnatal depression and children's growth and behaviour during the early years of life: exploring the interaction between physical and mental health
  1. Bilal Avan1,
  2. Linda M Richter2,
  3. Paul G Ramchandani1,
  4. Shane A Norris3,
  5. Alan Stein1
  1. 1Section of Child and Adolescent Psychiatry, University of Oxford, Oxford, UK
  2. 2Child Youth Family and Social Development (CYFSD), Human Sciences Research Council, Durban, South Africa
  3. 3Department of Paediatrics and Child Health, University of the Witwatersrand, Johannesburg, South Africa
  1. Correspondence to Paul Ramchandani, Department of Psychiatry, University of Oxford, Warneford Hospital, Headington, Oxford OX3 7JX, UK; paul.ramchandani{at}


Objective To assess the association between maternal postnatal depression and child behaviour problems and child growth at age 2 years

Methods This was a longitudinal birth cohort study in Johannesburg, South Africa. Primary analysis on the ‘Birth to Twenty’ cohort was performed for the association between maternal postnatal depression and child behaviour problems (n=1035) and growth (n=891) at age 2 and subgroup analyses (n=635) were carried out to assess the role of poor child growth in this association. Main outcome measures were the association between maternal postpartum depression (measured at 6 months postnatally using the Pitt depression inventory) and child behaviour problems (Richman child behaviour scale) and child growth at age 2 years.

Results Maternal postnatal depression was significantly associated with child behaviour problems at age 2, independent of socioeconomic status (β=0.353, p value=0.015). There was some evidence that children of depressed mothers were also at increased risk for having stunted growth, compared to non-depressed mothers (OR 1.61 (95% CI 1.02 to 2.56). The association between postnatal depression and child behavioural problems was significantly mediated by the stunted growth of the child (β=0.294, p value=0.111).

Conclusions Maternal postnatal depression is associated with later child behaviour problems independent of the socioeconomic status of the family. This association is mediated by the child's growth, demonstrating the importance of considering a child's physical and mental health together.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Clinically significant behaviour and emotional problems are very common throughout the world.1,,7 These problems include hyperactivity, conduct and emotional maladjustments and somatisation. Moreover, there can be long-term detrimental consequences of this untreated psychopathology at individual and societal levels; including reduced school attainment, difficulties in forming close relationships, adult mental health problems and an increased risk of psychosocial problems such as antisocial behaviour.8,,11 Consequently, the prevention of child behavioural and emotional problems should be a high priority and an improved understanding of the aetiology of these problems could potentially enable us to prevent child disorder and improve adult functioning.

What is already known on this topic

  • ▶. Postnatal depression is associated with an increased risk of behavioural and emotional problems in children.

  • ▶. Recent research from South Asia has suggested that children of postnatally depressed mothers may also be at increased risk for undernutrition and poor growth

  • ▶. There has been no research examining the confluence of these risks to physical and mental health in children, and limited research from developing nations, including those in Africa.

What this study adds

  • ▶. Maternal postnatal depression is associated with child behaviour problems and stunted growth at age 2 in a socioeconomically deprived, urban South African setting.

  • ▶. The association between postnatal depression and child behaviour problems is primarily mediated by poor child growth.

  • ▶. There is a strong independent association between concurrent undernutrition and behaviour problems of the child, irrespective of socioeconomic status.

Maternal mental health is intricately related to child physical and mental health.12 More recently maternal depression has been found to be associated with an increased risk of child malnutrition. This is mediated by impaired parenting ability, including difficulties in mother–infant interaction and to fulfil the infant nutritional needs and, in some cases, reduced ability to access appropriate healthcare resources. However, most of the compelling evidence for this association has been limited to the underprivileged regions of Asian countries.13 14 Similarly, the prevalence of behavioural and emotional problems is significantly higher among children of depressed mothers as compared to well mothers.15,,20 Maternal depression is associated with difficulties in mother–infant interaction as depression has been shown to interfere with the parents' capacity to provide adequate responsiveness and stimulation, factors that are critical in supporting early child emotional and cognitive development. Maternal depression is also associated with an increased risk of marital conflict and overall disruptive family functioning. These features may account in part for the increased risk of emotional and behavioural problems in their children.

Despite its clinical and social significance, some important dimensions of the associations between maternal depression and early child development are still underexplored. First, an underprivileged environment is a known risk factor for maternal postnatal depression and poor child development,21,,24 yet many research studies lack adequate assessment of the socioeconomic status. Second, child nutritional status or growth has not been considered in previous studies examining postnatal depression and child behaviour problems, despite the evidence that these are determinants of early childhood behaviour.25 26 This may be a particularly important consideration in developing country settings, where the majority of the morbidity associated with parental mental illness resides.

There has been relatively little research into childhood behaviour problems in developing countries, especially Africa. This is important because sociodemographic adversity has been shown to be an important risk for conduct problems. In this context the purpose of this paper is to determine whether there is an increased risk of behaviour problems and stunted growth at age 2 years in children whose mothers experienced significant depressive symptomatology in the postnatal period. We also aimed to examine the relationship between these physical and psychological components of child development.



The study is based on the ‘Birth to Twenty’ birth cohort, a longitudinal study based in Soweto–Johannesburg in South Africa.27 All singleton children born within a 7-week period during 1990 were enrolled in the study. The aim was to map the physical and psychosocial development of an urban cohort of children across the first two decades of life. This cohort of mothers were pregnant and delivered their babies during a extremely volatile time, politically and socially, in South Africa's history, and were exposed to high levels of political violence. Approximately 3200 women were originally recruited. Interviews were conducted by seven trained, multilingual, interviewers. Consensual agreement on the phrasing of questions was reached where different languages were required. The majority of interviews were conducted in the hospital clinic in Soweto, with a quarter conducted at home. Zulu, Sotho and English were the most common languages used. The present study uses data collected at assessments when the children were 6 and 24 months old. Ethical approval for the study was granted by the Committee for Research on Human subjects at the University of Witwatersrand, South Africa, and consent was obtained from all participating women.


Four groups of information were collected for these analyses, as outlined below.

Pitt depression inventory

Maternal postpartum anxiety and depression status were assessed at 6 months using the Pitt Inventory. It has 24 items and primarily assesses the current feelings and change in mood manifestations.28 Each item was answered and coded as ‘Yes’ (2), ‘Don't know’ (1) and ‘No’ (0). Total scores ranged from 0–48, and scores of 19 and above were considered as indication of postpartum depression. This measure has been used in a number of studies internationally, including in South Africa29,,31 and it correlates highly with other measures such as the Edinburgh Postnatal Depression Scale.32

Richman child behaviour scale

Children's behavioural difficulties were assessed using this 12-item questionnaire at age 2. The scales are completed by the parent and assesses a range of behavioural difficulties including, eating problems, sleep disturbances, soiling, hyperactivity, lack of concentration, poor relationships, tempers and fears.33 34 Some changes to the questionnaire were made to adapt it to the local culture. Total scores ranged from 0–12, and 42% of the children had scores that were designated as representing significant behaviour problems.

Anthropometric assessments of child

Height and weight were assessed at age 2 according to the standard procedures recommended by the World Health Organization (WHO).35 Electronic weighing scales were used to weigh the child to the nearest 0.1 kg and a stadiometer was used to assess height to the nearest 0.1 cm. The measurements were compared with WHO growth standards using WHO Anthro 2005.36 Three anthropometric indices were used: height for age (HAZ), weight for age (WAZ) and weight for height (WHZ), expressed in terms of z scores. These z scores were then used to form the three indicators: stunting (HAZ<–2), underweight (WAZ<–2) and wasting (WHZ<–2).

Demographic information

This was collected by maternal interview using a standard questionnaire in the perinatal period. Information collected included maternal ethnicity, age, education, household assets and housing quality.37 38

Statistical analysis

The analysis was performed in four main steps:

First, an analysis of possible attrition bias from the study was undertaken, comparing the main outcome measures in the followed-up sample and those who dropped out of the study.

Second, a socioeconomic index was created using principal components analysis (PCA), that is, on the basis of variance of the household assets, housing quality, house ownership and maternal education; standardised weights were assigned to them. These weighted scores were added by household to create the SE index. Using this index, study participants were placed into five equal socioeconomic groups with one corresponding to the lowest SE quintile and five representing to highest quintile of socioeconomic status.

Third, child behaviour scores were examined by a range of possible confounding variables, including socioeconomic status, ethnicity, birth weight, gestational age, maternal age and growth.

Fourth, univariate analyses were undertaken to examine for associations between maternal depression and childhood behavioural problems and growth. Multivariate analyses were then undertaken to adjust for any impact of socioeconomic status and other possible confounders.


Description of the sample

A total of 1860 women completed the Pitt depression inventory at 6 months postnatally. Information on Richman behaviour scale was collected at age 2 for 1737 children. However, data at both time points were available for 1035 (55.7%) and this comprised the sample for our primary analyses. Among this group of children data on growth status was available for 651 (63.0%), although overall data was available for maternal depression at 6 months and child growth status at age 2 for 1142 children. The original and follow-up samples are compared in table 1. The comparison revealed no significant differences between the follow-up and the attrition samples in terms of psychological assessments, nutritional status or socioeconomic status, suggesting that the study sample is broadly representative of those originally recruited.

Table 1

Comparison of those psychometric assessments those who participated in the analysis versus who were not

The majority of the studied sample was of black African origin (85%) followed by mixed ethnicity, whites and Asians (see table 2). About 91% of mothers were less then 35 years of age at the time of birth of the index child. Boys and girls are equally represented. Low birth weight and preterm delivery ranged from 10.4% to 12.2%. The socioeconomic characteristics of study sample were described in terms of household assets, housing quality, house ownership and maternal education as shown in table 2. Almost all the mothers have some level of literacy and the majority of the households had at least some modern amenities such as a toilet, water, electricity and a telephone line.

Table 2

Characteristics of study sample

Study measures

In all, 250 (24%) mothers had high scores on the depression inventory. One-third of the studied children had five or more concurrent behaviour problems at age 2. Using previously validated norms from UK based studies using the Richman Behaviour Scales, 10.5% of the children were rated as having significant behavioural problems. On anthropometric assessment at age 2, 16% of children were stunted, 4% were wasted and the prevalence of underweight was 3%.

Sociodemographic features and the main predictor and outcome variables

No statistically significant socioeconomical differential was observed for scores on the Pitt depression inventory (p value=0.07) and Richman scale (p value=0.78). In general, socioeconomic status showed a positive trend with the nutritional status that is, poorer children were more likely to be malnourished; however the association was significant for stunting only (p value≥0.001).

Boys were more likely to have more behaviour problems at age 2 years (see table 3). Among anthropometric indicators, stunting was significantly associated with the child behaviour problems (p value=0.039).

Table 3

Child behaviour scores at age 2 by child characteristics

Maternal postnatal depression and child behaviour and growth

Maternal postnatal depression was significantly associated with child behaviour problems at age 2 years (β=0.363; p value=0.011). This association held when controlling for child characteristics and socioeconomic status.

There was also stronger evidence of an association between maternal depression and child stunting at age 2 years, when adjusting for socioeconomic status (SES) index, maternal age, child gender and preterm delivery (see table 4). Children of depressed mothers were at 1.61 times higher risk for being stunted as compared to non-depressed mothers (p value=0.04).

Table 4

Regression analysis of maternal depression at 6 months with child anthropometric status at age 2

When the association between maternal postnatal depression and child behaviour at age 2 was examined controlling for the effect of child stunting, the association was no longer statistically significant (see table 5), suggesting that the association is mediated by the nutritional status of the child.

Table 5

Regression analysis of associated factors maternal depression at 6 months with child behaviour scores at age 2

Finally, additional regression analysis was performed on the subsample of mothers and children for which anthropometric data on stunting was not available. The significance and direction of association between maternal depression and child behaviour problems remained the same indicating the robustness of the association and lack of impact of attrition bias.


In this socioeconomically disadvantaged birth cohort drawn from Soweto, South Africa, we found that maternal postnatal depression was associated with child behaviour problems at age 2, independent of socioeconomic conditions. However, the most striking finding was the fact that this association was primarily mediated by nutritional status of the child. There is a strong independent association between current poor anthropometric status and child behaviour problems. Our study was the first study in African setting, which showed early maternal postnatal depression is directly linked to stunting at age 2.

Our study finding suggests that approximately 10.5% children at age 2 have significant behaviour problems. The general consensus is that child behaviour problems increase with age and potentially up to 15% of preschool children can have these problems39; however most of this data is from developed countries including UK (13%), Japan (10%) and the USA (11.8%) with relatively increased prevalence for African American children to 13%.4 40 41 In general there is a paucity of data from developing countries. Our study shows that there are similar trends in child behaviour problems at age 2 from this African study and the data from developed world. The high prevalence of this disorder highlights the need for active recognition of child behaviour problems by parents and service providers, and promotion of timely health seeking behaviour.42 The prevalence of the anthropometric indicators was significantly lower then the national average of South African children. The underestimation by our study could be explained by the fact that our studied sample is primarily representative of an urban population of South Africa.

Maternal depression

Inadequate maternal mental well-being is known to be directly related to child mental health including behaviour problems. Our study findings are consistent with the existing literature that children of depressed mothers are likely to show signs of greater behaviour difficulties.43 Maternal postnatal depression is a phenomenon of heterogeneous signs and symptoms. Certain components of depression can be relatively more detrimental and addressing those aspects can play a major role in management of behaviour problems of children of depressive mothers. We also suggest that clinical presentation of maternal depression could vary from culture to culture, and requires identification of particular components of maternal depression affecting child behaviour respective to community settings.

Child nutrition

Child malnutrition is a major public health concern in developing countries and maternal depression is a well established antecedent.14 44,,46 Similarly, our study demonstrated the links of maternal postnatal depression with poor nutritional status. However the mediatory role of child undernutrition has never been explored in the context of maternal depression and child behaviour problems during early childhood. One biologically plausible mechanism for this association could be that maternal depression impairs the mother's ability to feed her child properly irrespective of child socioeconomic status47 48; consequently the resultant undernutrition impairs the age appropriate maturity of neuromuscular and mental domains resulting in impaired child behaviour development. The strong influence of stunting highlights the influence of chronic undernutrition in behaviour problems during early childhood and stresses the need for in-depth expansion of this potentially important initial inquiry in future research.

Among other child characteristics, premature birth also contributed toward problematic child behaviour. This finding is well supported in the scientific literature and has been adequately adjusted in our analysis.49 50 52 53

The literature strongly supports the influence of maternal depression on behaviour problems and is more pronounced with child's gender. We found no gender contribution in effect modification on the association, which could be partially explained by cultural variation in child rearing norms and parental expectations.


Maternal ratings of child behaviour may potentially be affected by the mother's mental state, with depressed mothers tending to report a more negative pattern of child behaviour compared to mothers who are not depressed.43 We did not conduct direct observations of the child's behaviour as this was not feasible in a large community based cohort. However, maternal reports do provide a reliable observation of child behaviour51 54 as mothers are usually the primary care givers and the instrument used is a widely validated one that has been used in a wide range of contexts. We used a validated screening measure of depressive symptomatology but this does not provide a clinical diagnosis of depressive disorder syndrome. Furthermore, our study assessments were not supported by the clinical diagnoses of a partial sample, which could have added further validation to our results. A further limitation of the study was lack of repeated maternal depression and child nutritional status measurements. It is also of note that while stunting was clearly associated with later behavioural problems, other potential measures of malnutrition such as underweight and wasting were not. However, the findings for the association between behavioural problems and low birth weight and wasting at 2 years clearly showed the same direction of effect, although they were not statistically significant. Nonetheless it suggests that some caution should be exercised in evaluating these findings until further replications are reported in other studies.

Although this sample size was large, the response rates varied with different study measurements primarily due to inconsistent funding of the birth cohort and other operational challenges, but this did not appear to introduce any significant bias.

Chronic infections such as HIV and other conditions including iron deficiency and lead toxicity are known to be associated with generalised malnutrition and behaviour problems of children.55 56 However, assessment of these factors was beyond the scope of this study.

A difficult temperament in the child could also cause maternal stress. However the evidence shows that the direction of association between maternal depression and child behaviour is persistent irrespective of child's earlier behaviour and rules out reverse causality.


This was a large birth cohort study conducted in a developing country context and the prospective design enabled us to measure maternal depression and child behaviour in a chronological order. Another strength of the study is the detailed assessment of multiple socioeconomic factors, which have the potential to influence maternal and child well-being. Controlling for the child nutritional status has the noteworthy role in identifying the early association between maternal depression and child behaviour problems.

Study implications

The present study suggests that more than 10% preschool children are likely to present with behaviour problems at community level in a developing country setting. They should be assessed at health and education services level in the context of maternal mental health, socioeconomic conditions and child's nutritional status. It also reinforces the need for development of effective prevention and management interventions at community level, which could simultaneously address the mother and child mental health, and to delineate the strategies to provide nutritional and emotional support to the children affected by the maternal depression. However, we also emphasise that heterogeneity of maternal postnatal depression symptoms and specific cultural manifestations should always be considered while designing these interventions.


Maternal postnatal depression during early infancy appears to have an enduring negative effect on child behaviour; this association is mediated by nutritional status of the child at age 2 and the association is independent of socioeconomic conditions.


The authors gratefully acknowledge the contributions of the Birth to Twenty study team and the participants that enable this research to continue.


View Abstract


  • Funding The Birth to Twenty Research Programme is financially supported by the Wellcome Trust (UK, grant number 077210), Medical Research Council of South Africa, Human Sciences Research Council of South Africa and the University of the Witwatersrand, Johannesburg. PGR and AS are supported by the Wellcome Trust. BA, LMR and SAN report no biomedical financial interests.

  • Competing interest None.

  • Ethics approval This study was conducted with the approval of the Ethics committee of the University of the Witwatersrand.

  • Provenance and peer review Not commissioned; externally peer reviewed.