30 Depressed mothers and their children: attachment issues

Depression is a common psychological disorder in mothers. This is not only because female rates are nearly double male rates at different life stages (Nolen-Hoeksema 1990), but also because motherhood confers added risk (Bebbington et al. 1991). This results in dual social and psychological problems: first around the health and lack of well-being of mothers who suffer the depression, and second around the health and development of their children on whom it impacts. Thus women’s quality of life, intergenerational transmission of risk, and obstacles to a child’s healthy development become inter-related problems linked to maternal depression. Parenting and attachment style are both implicated.

Rates of major depression in inner-city working class mothers are quoted as 15% yearly, with 10% new onsets (Brown et al. 1990). This is similar to the 10–15% rates for women in the six-month postnatal period (Boyce 2003). Whilst depressive disorder can encompass bipolar and psychotic varieties, these are substantially less common than major depression (for example, psychotic depression affecting 1 in 1000 women after childbirth). Major depression is the focus of this chapter, defined by the Diagnostic Statistical Manual of Mental Disorders (DSM) as a series of symptoms, of which depressed mood or loss of interest are crucial, and others include weight loss or gain, insomnia/hypersomnia, agitation or retardation, fatigue or energy loss, feelings of worthlessness, diminished ability to think, concentrate, or make decision, and recurrent thoughts of death or suicide plans or attempts. Clinical level disorder is signified by the overlap of five such symptoms which need to occur, most of the time, for a minimum of two weeks with significant distress or impairment to functioning (American Psychiatric Association 1997).

Despite the high prevalence of such disorder, treatment for mothers is by no means easily available or accessed. For example it is estimated that 40–50% of episodes of postnatal depression go undetected in the United Kingdom (Seeley et al. 1996) with similar estimates for depression at other life stages (Paykel et al. 1997). This is due to a variety of reasons: stigma associated with depression, mistrust of medication, constraints around help-seeking, fear of having a child taken away, and high waiting lists for counselling services in primary care. Identifying the causes and correlates of maternal depression can aid in the public understanding of the disorder to help with detection and service-access.

Depression is a culmination of social adversity and psychological vulnerability compounded by recent stressors to provoke an onset (Brown et al. 1990). A hostile relationship with partner, lack of support from close others, and low self-esteem, constitute the proximal vulnerability factors which increase risk of disorder (Brown et al. 1990). This vulnerability is primed by more distal early life history of neglect and abuse in the woman, together with adverse adolescent and early adult experiences (Harris et al. 1987; Bifulco et al. 2000). Thus an episode of depression in the mother carries with it an accumulation of vulnerability and stressors and all can impact on her child.

Mechanisms by which maternal depression can have an impact on offspring are numerous. Those researched include parenting behaviour, attachment behaviour, and stress responses. Whilst genetic factors have also been identified (Caspi et al. 2003; Wilhelm et al. 2006), these are not as yet implicated in parenting behaviour. Depressive disorder can impede mother–child interactions and the quality of care provided in infancy (Murray et al. 2003), in later childhood (Belsky and Vondra 1989; Hammen 1992; Murray et al. 2003), in adolescence (Kestler and Lewis 2006), and in relation to child safeguarding services (Cleaver et al. 1999). However, studies increasingly show that other factors causally related to depression may have greater impact. The medicalization of risk transmission by identifying depression as the source of the problem, masks a more complex model which involves social and psychological life-course issues in mothers, transmitted to their offspring through parenting, attachment, and social adversity, including the behaviour of partners.

There have been a number of studies showing the negative impact of maternal postnatal depression on the mother–infant relationship. Thus depressed mothers’ engagement with their infants is characterized by hostile intrusive behaviour or withdrawal and disengagement (Cohn et al. 1990; Field et al. 1990). These maternal interactive profiles are accompanied by distinctive styles of infant disturbance, including distress and avoidance. Depressed mothers are less sensitively attuned to infants, less affirming, and more negating (Murray et al. 2003). However, the studies also show the impact of social factors: first the negative effects on infants are significantly associated with the mother’s life events and difficulties as well as her disorder (Murray et al. 2003). Second, among depressed populations who are not disadvantaged, extreme forms of infant disturbance do not occur (Field 1992).

In older children, the parenting task is identified in terms of the provision of adequate care and control, attunement to the child’s capabilities, and development. Poor parenting is determined by the quality of marital relationships, work, and social networks (Belsky and Vondra 1989), and these are influenced over the life course by prior developmental history and experiences. Stress factors emerge as more important than maternal disorder in relating to parenting outcomes (Shaw et al. 1994).

The marital or partner relationship is seen as key to parenting success. In mothers, conflict in the partner relationship is associated with critical parenting and with low emotional responsivity, and this together with powerlessness in the marital relationship lead to conduct problems in children (Webster Stratton and Hammond 1999). Similar findings are shown for fathers, but here marital powerlessness does not add. The quality of relationship with a partner is also central to models of women’s affective disorder. Negative effects on the women can occur through a number of mechanisms, such as the increased burden of sole parenting when partner is absent (Brown and Moran 1997), or through the impact of partners’ violence or other unpredictable behaviour related to alcohol use, antisocial disorder, and poor work history (Andrews and Brown 1988). It is also clear from examining accounts of childhood experience of the next generation that the absence of the father, family discord, violence involving the father, and abuse from the father are key for disorder in the offspring (Bifulco et al. 2009). The marital/partner relationship is a critical candidate in the transmission of risk simultaneously to mothers and offspring. This can also be protective—studies of young women with an institutional care background have better parenting behaviour if living with a socialized partner with no obvious psychopathology (Quinton et al. 1985).

Attachment theory provides a powerful framework for understanding and investigating parent–child interaction (Bowlby 1969, 1973, 1980). However, it has a wide remit and it also explains relationship patterns in general (for example, partner relating, support-access, and parenting capacity) as well as providing linkages with early life parenting experiences and risk of psychological disorder. It also illuminates the intergenerational transmission of risk. Attachment style has been investigated across the lifespan, including in perinatal investigation (Fonagy et al. 1991), parenting of children and adolescents (Shapiro and Levendosky 1999), and neglect/abuse of children (Crittenden 1997). It has also been examined in relation to the quality of marital/partner relationships (Hazan and Shaver 1994), accessing support (Larose and Bernier 2001), and mental health (Hammen et al. 1995; Sable 1997; Dozier et al. 1999; Muller and Lemieuz 2000). Thus within the attachment framework, parenting, partnerships, support, and depression are all implicated. All can be shown to converge on mother–child interactions.

The tenets of attachment theory hold that inconsistent, uncaring, or hostile parenting in the early years of life have long-lasting consequences for future relating style and adjustment. This occurs through ‘internal working models’ or cognitive structures developed in early years to create a template for future expectations of relationships (Fraley and Waller 1998). These working models are expressed in terms of secure or insecure styles of relating to others. Those with secure styles have a positive expectation of support, care, and warmth from those who are close to them. Those with insecure styles have expectations of harm, rejection, abandonment, betrayal, or unreliable support or care. Such styles have implications for the care of children intergenerationally, as experiences of poor parenting in early life can be repeated (van IJzendoorn 1995), but they also have other implications such as negatively influencing the choice of partner or support figures and failure to access emotional support (Rholes et al. 1998).

Attachment research on parenting has typically involved closely observed interactions between parent and infant in representative samples to examine normative experience (Ainsworth et al. 1978; de Wolff and van IJzendoorn 1997). However, the extension of attachment research to families living in more extreme social conditions, in relation to neglect, abuse, and maternal disorder, has moved the study of attachment further into the fields of child safeguarding (Crittenden 1988; Howe 2003) and psychopathology (Dozier et al. 1999). Insensitive parenting in the child’s early years may, under certain adverse conditions, become implicated in maltreatment further down the line. Thus, models of parenting that involve social deprivation, marital disharmony, parents’ work life, social isolation, and parents’ early life history are gradually becoming incorporated into the attachment approach (Belsky and Vondra 1989; Shaw et al. 1994).

Most of the research on attachment and parenting has utilized the Adult Attachment Interview (AAI) for parents, together with the Strange Situation test (SST) for assessing infant style. However, more contemporary instruments include the Attachment Style Interview (ASI) which is a support-based assessment of attachment style used for parents which has examined mothers in relation to parental support context, depression, and parenting of adolescent children. The two instruments and relevant research will be examined in turn.

The Adult Attachment Interview (George et al. 1984) is a well-established research tool for assessing attachment style in adults, based upon questions about childhood and early life experience, with discourse analysis of transcripts to ascertain cognitive features related to attachment such as coherence, idealization, derogation, and recall. These are used to derive attachment style categories in terms of autonomous–secure, dismissing–insecure, and preoccupied–insecure style in addition to ‘unresolved loss’ (Prior and Glaser 2006). The measure has been particularly orientated towards parenting style and the transmission of attachment patterns from mother to child. Thus, the AAI has been used in mother–infant pairs to examine similarity of style in the dyads, where attachment style in the infants is determined by the SST (Ainsworth et al. 1978). In this paradigm, the interaction within mother–child dyads plays out the expression of the internalized secure, anxious–ambivalent or avoidant working models of attachment in both participants (Main et al. 1985). The transmission of attachment insecurity across generations is argued to be through maternal insensitivity, intrusiveness or detached interaction, and poor emotional regulation when in contact with the child (de Wolff and van IJzendoorn 1997). This is indicated by significant concordances shown between mother and infant of around 65–72% (van IJzendoorn 1995). The consistency of mother–infant attachment categories also holds prospectively; for example, from maternal antenatal assessments of attachment style, to child attachment behaviour in the SST at three months (Fonagy et al. 1991), and later at 12–18 months of age (Steele et al. 1996). Reviews of the literature show a consistent association of maternal depression with insecurity in the child (Martins and Gaffan 2000) with lifetime maternal depression similarly predicting poor mother–infant interactions and insecure infant attachment (Carter et al. 2001).

Family systems approaches have added to the parent–child focus on intergenerational transmission by examining the contribution of either parents’ attachment style, as well as the effects of the relationship between parents, on children’s development and disorder. Studies by Cowan and colleagues have specifically explored the quality of the parents’ relationship as partners, their attachment styles, and problem behaviour in their preschool-age children (Cowan et al. 1996). In a sample of couples and their first-born child, the AAI was administered to both mothers and fathers, together with assessments of marital quality, parenting style, and child problem behaviour—both internalizing and externalizing. While no direct associations were found between parents’ attachment scores and the child’s problem behaviour, path analytic models showed different outcomes for mothers’ and fathers’ behaviour and for type of disorder outcome. The models showed that mothers’ attachment representations, quality of marital relationship, and parenting style provided the best fit for explaining internalizing behaviour in the child, while for fathers, attachment representations followed a similar model but for externalizing behaviour. For mothers, the quality of marriage was found to be independent of adult attachment status with positive marital functioning acting as a buffer, protecting the child against the negative impact of insecure attachment in one or both parents.

The social psychological approach to measurement of adult attachment style is built around support, and typically includes a set of questions about relationships with romantic partners and close support figures (Hazan and Shaver 1994; Mickelson et al. 1997). The Attachment Style Interview (ASI) is one of the few interviews measures that provides such contextual information while also coding attachment style categorizations (Bifulco et al. 2002a). This instrument assesses the quality of close relationships in adulthood (with partner and other adults named as very close) as well as attitudes indicative of both secure and insecure styles (e.g. Enmeshed, Fearful, Angry–Dismissive, and Withdrawn) (Bifulco et al. 2002b). Insecure attachment style has been shown to mediate between childhood neglect/abuse and onsets of depression in mothers (Bifulco et al. 2006a) with evidence emerging that parenting of the next generation is also implicated.

Findings similar to those reported by Cowan and colleagues concerning mothers were found in a study of intergenerational transmission to adolescent offspring. Mothers’ insecure attachment style was assessed with the ASI in relation to her history of partner relationships, her parenting competence, and emotional disorder in her adolescent offspring (Bifulco et al. 2009). The sample comprised 146 high-risk, mother–adolescent offspring pairs in London, who were recruited on the basis of the mothers’ psychosocial vulnerability for depression. Interviews were undertaken independently with mother and offspring. A path model was developed which showed that while mothers’ insecure attachment style had no direct link to her offspring’s report of maternal neglect/abuse, or to the young persons’ disorder, indirect links occurred through mothers’ incompetent parenting, partners’ problem behaviour and marital adversity. Mother’s lifetime depression, although common in this series, did not add to risk in her offspring.

The ASI has also been utilized across European and United States centres to assess its utility as a risk marker for maternal perinatal disorder (Bifulco et al. 2004). The study showed insecure attachment related to lower social class position, more negative social context, and major or minor depression both antenatally and postnatally. A specific association of Enmeshed or Fearful styles and postnatal disorder was found. Other findings from the same study showed poor partner support in pregnancy was related to poor mother–baby interaction at six months, with no effect from maternal depression (Gunning et al. 2004). Evidence is also emerging that insecure attachment style antenatally also related to the mothers’ insensitive interaction with the baby at six months (Bifulco et al. 2006b). A Portuguese study showed insecure attachment style using the ASI was also related to teenage pregnancy and depression during pregnancy (Figueirido et al. 2006). Pregnant teenagers were found to be nearly three times more likely to have an insecure attachment style than adults. Enmeshed style and poor partner support provided the best model for depression in both teenagers and comparison adult mothers.

These studies all indicate that an ecological approach (emphasizing social adversity and different role domains), and a lifespan approach (emphasizing a history of adverse relationships at different life stages) can be used in an attachment framework to indicate mechanisms of transmission of risk to the next generation.

The negative impact of maternal depression on infant health and development is increasingly shown to work through associated problems in relating, compounded by social adversity. These can be encompassed by attachment and social ecological approaches, to show effects on the children. This has important implications for preventative and early intervention work with families. The recent government agenda has highlighted children’s development and well-being in its Every Child Matters legislation (www everychildmatters.gov.uk). Whilst this approach is child-centred, the promotion of parenting programmes for mothers is also advocated, although the emphasis on the mothers themselves and their psychosocial risks receives less attention.

The use of the ASI or similar assessments for health visiting services, or to use alongside parenting assessments, may prove useful for intervention work. This is already happening in the adoption–fostering field to gauge carer vulnerability or resilience (Bifulco et al. 2008). Given health and care services working jointly with education to focus on children in schools, this may also be an opportunity to capture mothers vulnerable to disorder in these contexts for intervention. Thus parenting programmes could be geared not only to care provision and mother–child interaction, but also to maternal help for depression and associated vulnerability. Research needs to examine whether parenting programmes may serve not only to improve mother–child interaction, but also improve mother’s psychosocial risk and disorder.

Depression is a very treatable disorder, with both therapeutic approaches and medications shown to have significant beneficial effects, whether treated in general practice (Schulberg et al. 1995; Rost et al. 2000) or in the postnatal period (O’Hara et al. 2000; Stuart et al. 2003). Problems arise when depression is undetected and access to services restricted. Whilst screenings in the antenatal period are common, these can miss psychosocial vulnerability. Depression is known to be increasing with much still undetected, despite its causes being known and treatments available. It is perhaps partly due to the scale of the problem that there is insufficient proactive work in the area. The Layard report has indicated the large size of the cognitive behavioural therapy workforce required to tackle the scale of the problem (Layard 2006). Whether treatments for depression will also tackle the underlying psychosocial risks which affect parenting and children’s development is unclear. Both aspects need to be tackled in order to improve well-being in families.

Problems in parenting and family life need to be seen in the context of social adversity. Disadvantage, stress, failed relationships, and loss of support are all key to understanding negative impact on children and mothers. Intervention is important at different points, both in the child’s development—infancy, mid-childhood, and adolescence—but also at different stages in the development of maternal risk. Tackling maternal depression is rather late in the evolution of these problems. Whilst depression itself must be treated, earlier solutions for resolving problems would ensure less suffering and cascading of risk factors. This should be the aim of multiple agencies working with children and families.