Postnatal depression is associated with detrimental life-long and multi-generational impacts on relationship quality

Introduction

Postnatal depression (PND), functionally defined as a major depressive episode occurring within 12 months of giving birth (Stowe, Hostetter & Newport, 2005; Halbreich & Karkun, 2006; Skalkidou et al., 2012), has received a great deal of attention from psychologists due to its association with a range of detrimental outcomes in children (Gelfand & Teti, 1990; Beck, 1998). For instance, maternal PND has been found to correlate with lower infant weight gain and weight faltering during the first four months after birth in the offspring whose birth is associated with the depression (Wright, Parkinson & Drewett, 2006), poorer infant motor development by 15 months (Cornish et al., 2005), stunted growth at two years (Avan et al., 2010), and higher rates of morbidity (when monitored for the first four years) due to gastrointestinal and lower respiratory tract infections (Ban et al., 2010). Additionally, chronic depression in mothers is associated with poorer cardiac function in their children aged 9.5 years (Gump et al., 2009). PND is also implicated in poorer child behavioural and cognitive outcomes, correlating with, for example, behavioural problems at 18 months (Murray, 1992), reduced interpersonal functioning at 19 months (Stein et al., 1991), behavioural problems at 2 years (Avan et al., 2010), and lower cognitive ability at age four (Cogill et al., 1986). A review of the quantitative and qualitative literature also points to consistent findings of correlations with poorer language and intelligence development, as measured by intelligence quotient (IQ), throughout childhood (O’Hara & McCabe, 2013). PND is thought to have a disruptive impact on mother-infant bonding and attachment (Beck, 1995; Murray et al., 1996; Coyl, Roggman & Newland, 2002; Moehler et al., 2006); postnatally depressed mothers show heightened self-focus (Salmela-Aro et al., 2001), and are more likely to exhibit a range of potentially problematic parenting behaviours (for a review see O’Hara & McCabe, 2013). The negative effect on parenting is suggested to mediate the relationship between PND and subsequent child development, potentially alongside shared environmental or genetic factors influencing both PND aetiology and child development outcomes (Murray & Cooper, 1997).

The number of long-term studies on the impact of PND are small but growing (Sanger et al., 2015) and research now extends to the adolescent period; for example, PND experienced by the mother is correlated with an increased incidence of teenage depressive and anxiety disorders (Halligan et al., 2007; Murray et al., 2011), higher cortisol levels (Halligan et al., 2004), lower IQ (Hay et al., 2008), and lower academic attainment at age 16 in the child (Murray et al., 2010; Pearson et al., 2016). PND has also been associated with increased rates of marital discord and depression in partners (Boath, Pryce & Cox, 1998; Burke, 2003), which may further affect developmental outcomes in children from families where these issues arise.

However, little is known about the impact of PND on the longer-term relationships between mothers and their children. Lower mother–offspring relationship quality is linked to reduced offspring self-esteem and social competence (Kim & Cicchetti, 2004), which has potential consequences such as being less able to achieve high social status (Rudolph, Hammen & Burge, 1995) or succeed in the job market (Baron & Markman, 2003). Such effects are of theoretical interest from an evolutionary perspective as they indicate that mother–offspring relationship quality plays a role in the accrual of offspring embodied capital and hence offspring quality. Embodied capital refers to the attributes and resources an individual has available for conversion into biological fitness enhancing commodities (Kaplan et al., 1995), i.e., the attributes and resources that influence whether an individual succeeds in attracting a mate, and the resources available for investing in resulting children. Parents are thought to make investments in their children’s embodied capital based on their own capital and the perceived likelihood that their investments will pay off in the form of grandchildren. Therefore, mother–child relationships are of interest in both terms of their effect on offspring quality and offspring’s capacity to subsequently invest in their own children.

The influence of PND on intergenerational relationships is also of evolutionary interest, yet is currently unknown. Grandmothers are thought to play an important role in human evolution; evolutionary anthropologists have repeatedly highlighted the importance of grandmothers in enhancing child survival (Sear & Mace, 2008) and their influence on the reproductive trade-offs of their offspring (Hawkes & Coxworth, 2013) in pre-industrial contexts. A small body of literature also indicates that grandparents can enhance grandchild development and growth via socio-emotional support in low-risk industrialised settings (Coall & Hertwig, 2010). In Western contexts, it has been demonstrated that relationships with offspring mediate grandparent-grandchild relations (Michalski & Shackelford, 2005), positively correlating with grandparental involvement (a form on investment) with their grandchildren (Barnett et al., 2010). If PND negatively affects mother–offspring relationship quality, it is also likely to negatively affect grandmaternal investment from the grandmother who experienced PND via reduced grandmother–grandchild relationship quality. Grandparents make investments that influence their children’s reproductive and parental investment decisions even in modern, developed settings (Coall & Hertwig, 2010; Coall & Hertwig, 2011). For instance, in the UK contact frequency with grandparents predicts parents’ continuation of childbearing (Tanskanen et al., 2014) and the educational achievement of their children in the first year of school (Tanskanen & Danielsbacka, 2016; Tanskanen, 2017), contact frequency with grandmothers in the UK is associated with lower levels of breastfeeding in mothers (Emmott & Mace, 2015), and, in Bulgaria, supportive residential grandparents were found to protect adolescents from external stressors which otherwise correlated with increased depressive symptoms (Botcheva & Feldman, 2004).

Here we present results from a retrospective survey study, designed to gather the complete reproductive histories of postmenopausal women (generation 1), along with measures of their history of PND, depressive tendencies more generally, demographic characteristics, the reproductive histories of their offspring (generation 2) so far, and the relationship quality they have with their children (generation 2) and grandchildren (generation 3). The ages of the children of respondents (generation 2) ranged from 8 to 48 years, with an average age of 29 years, allowing assessment of the impact of PND over a longer period, to the best of our knowledge, than previous studies. We first test the hypothesis that (1) Mothers (generation 1) who experienced PND will have lower quality relationships throughout their lives with the children (generation 2) with whom the depression was experienced and then, by extension, (2) grandmother–grandchild relationship quality will be lower when the grandchildren (generation 3) come from a child (generation 2) whose birth was associated with PND.

Materials and Methods

Sample characteristics

Respondents (generation 1) had a mean age of 59.9 years (SD 7.4) and were born between 1930 and 1967 (Fig. 1). The majority of respondents did their childrearing in the UK (74.5%), followed by North America (12.9%), and were of high to medium SES (‘professional’ 67.9%, ‘managerial and technical’ 20.7%). Respondents gave birth to a mean of 2.2 children (SD 0.8, range 1–6). An actual diagnosis of PND, at at least one birth, was received by 40 women (13.1%), while 72 women (23.6%) met the criteria for PND according to the BPDS at least once, and 110 women (36.1%) met the cut-off for PND according to the EPDS at least once. Data on mother–child relationship quality was reported for 666 offspring (generation 2) (for 15 offspring there was non-systematic missing data); the mean PAI score was 42.8 (SD 5.6, range 17–50) and the mean age of these offspring was 29.1 years (SD 9.7, range 8–48 years). 125 respondents (generation 1) had 373 grandchildren (generation 3) born to 197 children (generation 2).

Results

(1) Mothers (generation 1) who experienced PND will have lower quality relationships throughout their lives with the children (generation 2) with whom the depression was experienced

1.i mother–child relationship quality (PAI score) was statistically significantly lower when the child’s birth (generation 2) was associated with PND (median = 43.5, n = 564) compared to when the child’s birth was PND free (median = 41.5, n = 102): W = 33, 656, p = 0.006, r =  − 0.106. The distribution of PAI scores by PND incidence can be seen in Fig. 2).

1.ii As PND severity (EPDS score) increased mother–child relationship quality (PAI score) decreased across age ranges (Fig. 3). Multilevel analysis showed this relationship remained after controlling for both individual effects of the mothers (generation 1) alone and with other factors (Table 1); a woman’s (generation 1) PND symptom severity at a given birth was negatively correlated with her relationship quality with the child from that birth (generation 2). The PND only model found that controlling for only PND severity accounted for 4% of the variance in mother–child relationship quality.

The following factors, which did not vary between a mother’s (generation 1) births, also negatively correlated with mother–child relationship quality (Table 1): having a general tendency towards DAS, and having medium to low levels of support from one’s own mother compared to having high levels of support from her. The age of the child (generation 2) had no impact on the relationship quality between mother (generation 1) and child (generation 2), nor did it interact with PND symptom severity (see R code in the Supplemental Information 1).

(2) Grandmother–grandchild relationship quality will be lower when the grandchildren (generation 3) come from a child (generation 2) whose birth was associated with PND

2.1 The relationship a mother (generation 1) has with her child (generation 2) was less likely to have a positive impact on her relationship with her grandchildren (generation 3) from that child if the child’s birth (generation 2) was associated with PND (Table 2): Fisher’s exact test = 5.227, p = 0.064 (two-tailed), Cramér’s V = 0.133.

2.ii Relationships with grandchildren (generation 3) were of lower emotional closeness when the birth of the grandchild’s parent (generation 2) was associated with PND (Table 2): Fisher’s exact test = 9.845, p = 0.006 (two-tailed), Cramér’s V = 0.222.

Discussion

The detrimental consequences of PND on mother–child relationships over the short-term are well documented, but until now little was known about any of the longer term impacts. This is the first piece of research to document that PND is associated with a reduction of mother–child relationship quality into the offspring’s adult years. Maternal (generation 1) self-reported mother–child relationship quality decreased as PND symptom severity increased, after controlling for a variety of other factors likely to influence relationship quality. The level of support from maternal grandmothers (generation 0), when alive during their daughter’s (generation 1) childbearing years, also positively correlated with mother–child relationship quality. Additionally intergenerational relationships also appear to be affected, with grandmaternal (generation 1) self-reported grandmother–grandchild relations more likely to be negatively categorised if PND occurred at the birth of the parent (generation 2), indicating a hitherto unknown consequence of PND.

Until recently, widespread screening for PND, in the UK at least, was considered to not be cost effective (Paulden, Palmer & Hewitt, 2009); however, there have been recent political moves in the UK to improve spending on maternal mental services (Campbell, 2016), and in 2016 the US Preventative Services Taskforce recommended population wide screening for depression in pregnant and postpartum women (O’Connor et al., 2016). Results pointing to the long-term impact of PND add impetus to the necessity of investing in preventative strategies, rather than simply identifying women once they are already depressed. A wide range of risk factors for PND are well documented (for a review see Yim et al., 2015) and successful intervention strategies for high-risk groups do exist (Austin et al., 2008; Dennis et al., 2009).

Most crucially, our results also suggest that while maternal tendency towards depression, anxiety, and stress, more generally across the life course reduces mother–child relationship quality, the negative effect of PND is specific to the child whose birth the PND was associated with, rather than being related to a lowering of relationship quality with all of a mother’s children. A proximate level explanation for this may be that the memory of PND has a permanent detrimental impact on relations with the child with whom it was associated. However, an evolutionary perspective suggests a potential ultimate level explanation: if mother–child relationship quality can be taken as a proxy for maternal investment, or indeed as a form of investment, this suggests that factors in early infancy may affect the pay-offs of maternal investment in terms of offspring quality which set the course of life-long investment levels.

Mother–child and grandmother–grandchild relationship quality are also likely to influence maternal/grandmaternal subsidy of offspring parenting costs, in turn impacting the investment received by grandchildren. Respondents (generation 1) were more likely, at a level approaching significance, to rate their relationship with their child (generation 2) as having a negative effect on their relationship with their grandchildren (generation 3) from that child if they had PND in association with the child’s birth. They were also more likely to report feeling lower emotional closeness with grandchildren (generation 3) when PND occurred in association with the birth of the parent (generation 2) of the grandchildren. This effect is likely to be mediated by the reduced mother–child relationship quality experienced in association with PND (Michalski & Shackelford, 2005); in the United States at least, lower quality grandmother-parent relationships predict lower levels of grandmother investment both in terms of contact frequency with grandchildren and help raising them (Barnett et al., 2010).

There is a large sociological and public health literature highlighting the importance of maternal grandmothers in helping their daughters raise their offspring (for a review see Scelza, 2011). Evolutionary anthropologists have also highlighted, largely in natural fertility populations, the importance of the support of grandmothers for maternal investment and subsequent child outcomes, with the presence of maternal grandmothers generally improving child survival rates (Sear & Mace, 2008). Thus, our finding that mothers (generation 1) who report higher levels of support from their own mothers (generation 0) tend to report higher quality relationships with their children (generation 2) is unsurprising.

It must be noted that the size of the effects documented here are small; however, given more immediate events and conditions are to be expected to have a large effect on the reporting of relationship quality, this is unsurprising. Nonetheless, these results indicate that conditions early in a mother and child’s relationship may have long-lasting consequences.

Conclusions

PND appears to have long-term detrimental consequences for mother–child relationships well beyond adolescence. Inter-generational effects of PND were also indicated, with PND not only associated with reduced emotional closeness with children but with the grandchildren of these children as well. As such, these results indicate that PND is implicated in a cascade of negative multigenerational effects and suggest that investment in PND preventative measures may not only improve relationships between mothers and their children in the short-term, but will also improve family relationships, closeness, and support well into the future.

Supplemental Information