Family Alliance as a Moderator of the Link Between Maternal Postpartum Depression and Child Symptoms Assessed by Both Parents

Abstract

We investigated the moderating effect of family relationships on the links between maternal postpartum depression and child symptoms in a low-risk community sample of families with 3-month-old infants (n = 57). The level of maternal depression was assessed by the Montgomery-Åsberg Depression Rating Scale from a clinical interview, child symptoms by the Symptom Check List completed by both parents, and family relationships by direct observation of father–mother–baby interactions (Lausanne Trilogue Play). Families were categorized as high coordination or low coordination from their overall coordination level throughout the play. Results showed no significant links between maternal depression level and child symptoms reported by both parents. Mothers with a high depressive level in high coordination families tended to report more symptoms in their child than did mothers with lower depressive scores, whereas this link was not found in low coordination families. Prevention perspectives and clinical implications of these results are discussed.

Keywords

child symptoms family alliance postpartum depression triadic play

Depression is a rather common disorder during the postpartum period; epidemiological data show that 8% to 20% of women develop a clinically significant depressive state during the first year following a child’s birth (Manzano, Righetti-Veltema, & Conne Perreard, 1997; Sheeder, Kabir, & Stafford, 2009). The incidence of maternal depressive symptoms in this period is up to three times higher than in any other period of life. Postpartum depression (PPD) has emerged as a specific diagnosis in the International Classification of Diseases as part of the mild mental and behavioral disorders associated with the puerperium (F.53.0; World Health Organization, 1992) and as a subcategory of depressive disorders in the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR; American Psychological Association, 2000). PPD shares most of its clinical features with major depressive disorder, such as low mood, fatigue, sleeping and eating problems, excessive and inappropriate guilt, and slow psychomotor functioning (O’Hara, 1997). However, the clinical descriptions also report parent-related symptoms in postpartum depressive mothers, for example, feelings of being inadequate in their maternal role or unable to take care of the child, excessive anxiety, or specific fears concerning the child’s health (Pitt, 1968; Ross, Evans, Sellers, & Romach, 2003).

During the last three decades of the 20th century, PPD has been found to seriously alter maternal behaviors and capacities that the mother needs to establish a positive relationship with the baby. For example, sensitivity and responsiveness to the child’s signals are reduced and intuitive parenting behaviors less adapted (Murray & Cooper, 1997a; Papousek & Papousek, 1997). PPD also negatively influences perceptions, as depressed mothers are more likely than nondepressed mothers to describe their infant negatively (such as having a difficult temperament; McGrath, Records, & Rice, 2008). In turn, Maternal negative perceptions about the child were shown to be associated with extended detrimental effects on the child along the infancy (Leckman-Westin, Cohen, & Stueve, 2009).

As most studies have compared groups of severely depressed (e.g., hospitalized mothers) versus nondepressed mothers, a body of evidence exists about the severe forms of this disorder and its consequences; in contrast, very few studies have focused on mild (or subclinical or subthreshold) forms of PPD in the general population. Although subclinical depressions may deteriorate mothers’ psychosocial functioning (Weinberg et al., 2001), little is known about the potential impact on the child. PPD has also been shown to be globally underdiagnosed (Knudson-Martin & Silverstein, 2009) and often not treated (Dennis & Chung-Lee, 2006), reinforcing the need to investigate less severe forms of depression in the postpartum period. One aim of this study was to screen for PPD in a community sample and to evaluate the impact of the level of maternal depression on the child in the early months of life.

Maternal Depression and Child Outcomes

To clarify the processes at work in these relational disturbances, numerous studies have focused on depressed mothers’ behaviors when interacting with their child (for a recent review about the numerous impacts of PPD on parenting and parent–child interactions, see Field, 2010). Reduced emotional availability caused by depression was shown to affect the maternal capacity to adjust to her child’s behavior, communicative signals, and emotional states in the interactions (Vliegen, Luyten, & Biringen, 2009). Whereas most depressive mothers display withdrawn behaviors, with poorer vocal and facial expression of emotions, as well as less physical and gaze contacts, others show intrusive behaviors, interfering with self-regulatory behaviors and overstimulating the infant (Field, Hernandez-Reif, & Diego, 2006). These behaviors were found to induce atypical child behaviors in the social exchange. When interacting with their depressed mother, children show low levels of positive emotions and high levels of negative emotions, vocal and visual withdrawal, overadapted reactions to intrusive stimulations, and constant alertness toward their mothers’ behaviors (Beebe et al., 2008; Tronick & Weinberg, 1997). They tend to be less active agent in the interactions, initiating less positive connections with their mother than children of nondepressed mothers (Dix, Cheng, & Day, 2009).

Long-term perturbations in child development include insecure attachment and behavioral problems due to emotion, arousal, or attention regulation difficulties (Cicchetti, Rogosch, & Toth, 1998; Hay, 1997; Lyons-Ruth, Lyubchik, Wolfe, & Bronfman, 2002; Tronick & Gianino, 1986). Cognitive development impairments are also observed, as maternal depression has been linked with early perceptive learning delays (Murray, 1992) or lower IQ scores at the age of 4 years (Cogill, Caplan, Alexandra, Robson, & Kumar, 1986).

These findings suggest that PPD constitutes a major risk for children for developing different psychopathological and cognitive disturbances during early childhood and infancy.

Family Relationships as a Moderator

Numerous researchers, however, have challenged the linear causal relationship between PPD and negative child outcomes. They underlined the moderating role of the psychosocial context, the characteristics of which may act as risk versus protective factors (Tharner et al., 2011). PPD seems to have a stronger impact on children in high-risk samples, which highlights the moderating effect of a set of variables, such as mothers’ socioeconomic status, education, social support, income, level of social stress, sense of self-efficacy, or marital status (Goodman & Brumley, 1990; Hay & Kumar, 1995; O’Hara, 1994; Weinberg, Olson, Beeghly, & Tronick, 2008). Beyond these maternal characteristics, relational and family-related variables were also found to play a moderating role. In two-parent families, the negative impact of PPD on child development seems to be exacerbated when associated with low marital satisfaction in mothers or marital conflict (Carro, Grant, Gotlib, & Compas, 1993; Goodman, Brogan, Lynch, & Fielding, 1993; McElwain & Volling, 1999). The presence of a nondepressed father has also been identified as a possible buffer, as he may compensate the difficulties of a depressive mother in being more sensitive and adjusted to the child in daily care (Edhborg, Lundh, Seimyr, & Widström, 2003) or may support the mother in her role (Bost, Cox, Burchinal, & Payne, 2002). However, some studies pointed out that this buffer effect of paternal involvement depends on its acceptance by the mother and its integration in her own relationship with the child (Teichman & Teichman, 1990). The father’s involvement may indeed be perceived by the mother as supportive or, on the contrary, as a confirmation of her own difficulties in the relationship with the child, reinforcing her feelings of inadequacy or inefficacy as a mother; these differences of perception again depend on the quality of the marital relationship. In the presence of conflict between spouses, the father’s involvement tends to be reduced (Paulson, Dauber, & Leiferman, 2011), and his engagement may possibly be felt by the mother as a threat. In contrast, a positive marital relationship may foster the paternal desire to be involved and may lead the mother to accept this involvement more easily, whether in the daily care of the child or as a support in her relationship with the child (Gotlib & Beach, 1995).

To date, few studies about PPD have focused on family-related variables. Although these questions are discussed in theoretical reviews or clinical papers, few empirical data exist regarding family-related factors and their possible moderating effect on maternal depression and its impact on the child.

Family Alliance

We used the concept of family alliance to investigate the processes occurring in triadic mother–father–infant interactions. These processes can be assessed by observing family interactions during a play situation, such as in the Lausanne Trilogue Play (LTP; Corboz-Warnery, Fivaz-Depeursinge, Gertsch Bettens, & Favez, 1993). Family alliance refers to the degree of coordination that father, mother, and child reach in completing a task (Fivaz-Depeursinge & Corboz-Warnery, 1999). In the family alliance model, four functions must be fulfilled to reach an alliance with a high level of coordination: each member must be included (participation function); each member’s role in the game must be defined to reach a sense of “who does what” (organization function); a focus must be shared to create a joint activity (focalization function); and this focus must allow intersubjective sharing as a condition for experiencing positive affects (affect sharing function; see Frascarolo, Favez, Carneiro, & Fivaz-Depeursinge, 2004). The fulfillment of these four interactive functions determines the quality of the family alliance: The more functions are fulfilled, the higher the coordination, and thus the quality of the alliance. The model assumes that failures in the two first functions (e.g., withdrawal or exclusion behaviors, difficulties organizing and respecting each other’s role, difficulties cooperating and working as a team) indicate predominant signs of miscoordination; the term low coordination alliance reflects the low quality of the interactions and the weak coordination between family members when playing as a family triad. In contrast, a family that is able to create joint activities and to have fun (third and fourth functions) shows more coordination to reach this goal; the term high coordination alliance refers, therefore, to the high coordination capacities of family members. The relevance of distinguishing these types of functioning has been validated in clinical descriptions and confirmed by cluster analyses (Favez, Frascarolo, & Fivaz-Depeursinge, 2006; Fivaz-Depeursinge, Frascarolo, Lopes, Dimitrova, & Favez, 2007).

Referring to the structural family systems theory developed by Minuchin (1974), the family alliance theory assumes that the family structure consists of patterns of interactions developed by family members in everyday life. In primiparous families, these patterns were found to take root during pregnancy and to remain stable in the child’s first 2 years of life (Favez, Frascarolo, Carneiro, et al., 2006), suggesting that the prenatal marital relationship is the matrix of postnatal family relationships (Favez, Frascarolo, & Fivaz-Depeursinge, 2006). Furthermore, data showed that a control group of nonreferred families from a community sample has a higher rate of high coordination alliances than does a clinical group of families referred for child psychofunctional disturbances (Favez, Lavanchy Scaiola, Tissot, Darwiche, & Frascarolo, 2011; Tissot, 2009).

Aim and Hypotheses

In this study, we addressed the question of links between maternal PPD and the presence of symptoms in the child at 3 months postpartum, as well as the moderating impact of family alliance on these links. First, we hypothesized that the level of depression in mothers would be linked to negative child outcomes. Second, we assumed that the quality of the family alliance would influence the strength of the association between PPD and negative outcomes in 3-month-old children. In other words, in families with a low-quality alliance, PPD was expected to be strongly linked to negative child outcomes. In contrast, a weaker link or no link was expected between PPD and child outcomes in families with a high-quality alliance. The study focused on a community sample to increase our knowledge about subthreshold PPD and its consequences in families in the general population.

Method

Sample

The data were collected between January 2009 and February 2011 from 65 two-parent families from a community sample, who participated in a larger study on the development of early family relationships. Maternal PPD and its influence on early child development were on purpose not defined as an explicit object of study. Families were recruited with various methods: at Babyplanet, an annual fair about child care; at the maternity service of the Lausanne University Hospital; and by means of flyers distributed during parental visits to the General Register Office in Lausanne, Switzerland.

Mean ages were 32.3 years (SD = 4.2) for mothers and 34.7 (SD = 5.7) for fathers. Socioeconomic status ranged from lower to upper middle according to Hollingshead’s classification (Hollingshead, 1975), with a median range in the upper-middle class for both mothers and fathers. In Hollingshead’s classification, the socioeconomic status is derived from the product of an educational scale and an occupational scale, leading to a 5-level classification (from “lower” to “upper”). Of the parental couples, 74% were married. All families were living near Lausanne, in the French-speaking part of Switzerland.

The children were all healthy and born at full term, with a distribution of 34 boys (52.3%) and 31 girls (47.7%). The mean age for children was 98.7 days (SD = 9.5). Most of them were first-born babies (67.2%).

The inclusion criteria were as follows: both parents and the infant had to live in the same household; families had to be fluent in French, as the questionnaires were administered in French; and the infant had to have been born without any identified neurological impairment. As the purpose of this study was to investigate the links between our variables in the community, we had to limit the inclusion and exclusion criteria to improve external validity.

Procedure

We received the families in the laboratory of the Center for Family Studies, Lausanne University Hospital, where they were asked to play in a triadic play setting. At the end of this first session, each family received a set of questionnaires. During a second session that took place at the family’s home, a trained psychologist assessed the presence and severity of depressive symptoms in the mothers by means of a semistructured interview, whereas another psychologist conducted an evaluation of child development.

Family Alliance

Families were invited to take part in the LTP, a validated observational situation designed to assess the quality of family interactions (Corboz-Warnery et al., 1993). In the LTP, seats are placed to form a triangular area at a distance fostering interaction. At 3 months, the infant sits in a baby chair, which can be oriented in three positions: toward one parent, toward the other, and between the two of them. The play is structured in four parts according to a scenario, which allows observation of all the possible ways for the three family members to interact as a triad: (a) one parent plays with the child (the active parent role), the other one being “simply present” (the participant-observer role); (b) parents switch roles; (c) the three play together; and (d) parents have a discussion, leaving the baby on her own for a while.

Each part lasts 2 minutes, with an investigator signaling the transition to the next part; 10 seconds are allowed for the transition between parts. This timing has been set according to the “naturalistic” duration of the classical LTP in less standardized procedures (Favez, Frascarolo, & Fivaz-Depeursinge, 2006). The order of parts “a” to “c” was counterbalanced to rule out an order effect. For coding purposes, the LTP was videotaped in a multiple-camera technical setting.

A coder who was not present during this first session was then asked to code LTP videotapes by using the Family Alliance Assessment Scales (FAAS), a validated coding system designed to assess the quality of triadic and family interactions during the LTP (for the validation study, see Favez et al., 2011). This tool contains prototypical descriptions of behavioral clues (see the descriptions below). The presence—or absence—of each clue helps the coder to determine the quality of the family alliance itself, which is the degree of coordination a family can reach in completing a task (Fivaz-Depeursinge & Corboz-Warnery, 1999). We then classified the families according to the two patterns of coordination.

In high coordination (HC) families, the members are all included in the situations and all participate in the play. The parents respect their own roles, as well as those of their spouse, according to the LTP scenario. They share a focus of interest during the play, each member working to enrich the game. Finally, this HC allows them to share positive affects during the play. In some HC families, however, the stress induced by the laboratory situation can make it difficult for them to fully experience positive affect sharing, even though their coordination is high enough to be classified as HC.

In low coordination (LC) families, several behaviors prevent the family members from coordinating to achieve the task. During the play, a member can exclude himself or herself or can be excluded by the others who do not pay any attention to him or her. Parents do not respect their role and those of the others, for example, interfering in the others’ play when they are supposed to be in the participant-observer role. This disrespect regarding the roles in each part makes the scenario disorganized. We sometimes observe parental competition to get the child’s attention, leading to overstimulation of the child who is then entrapped in the parental conflict. Generally, these behaviors prevent the family from reaching moments of positive affect sharing, and affects are mostly negative. The emotional climate can be felt as not authentic by the coding observer.

To assess the interrater reliability of the family alliance scale, a second rater coded 15 LTP videotapes randomly selected in the sample. The reliability between the two coders was considered to be good enough, with an ICC_2,1 = .75 for family alliances.

Child Outcomes

Parents completed a questionnaire about their children’s psychofunctional symptoms, the Symptom Check List (SCL; see Robert-Tissot et al., 1989, for the publication of the first version; the unpublished revised version F-95 has been used for this study). Parents also rated ten 5-point Likert-type scales concerning sleep, eating, behavior, and separation problems. For nine items, the parent rated the occurrence of a potentially problematic behavior of the child from never to often or from not characteristic to very characteristic. One item (item 4: eating behaviors) was rated on a bipolar scale from 1 (does not eat enough) to 5 (eats too much), both extremes being problematic. This item was thus not included in subsequent analyses.

This checklist has good predictive validity when comparing a nonreferred group of families with children up to 3 years old with families referred for psychofunctional symptoms (Robert-Tissot, Rusconi-Serpa, de Muralt, Stern, & Cramer, 1991).

Maternal Depression

Mothers were interviewed at home according to the Diagnostic Interview for Genetic Studies (Nurnberg et al., 1994; Preisig, Fenton, Matthey, Berney, & Ferrero, 1999), and the presence and severity of depressive symptoms were assessed with the French version of the Montgomery and Asberg Depression Rating Scale (MADRS; Montgomery & Asberg, 1979; for the French translation, see Lemperière et al., 1984). The MADRS consists of ten 7-point scales exploring depressive symptoms. Mothers were questioned about their feelings and mood during the whole postpartum period. The total scores range from 0 to 60 (α = .76), with cutoff scores at 9, 18, and 35 points (for mild, moderate, and severe depression, respectively; Müller, Szegedi, Wetzel, & Benkert, 2000; note that the 9-point cutoff will next be used as the criterion defining depressed vs. nondepressed mothers).

These interviews and assessment were performed by trained clinical researchers belonging to the epidemiology unit of the hospital, which is independent of our center. To preserve neutral coding of the interactions in the LTP, we did not receive these data until the end of the family interactions coding process.

Control Variable

To ensure that the potential symptoms of the children were not due to any developmental delay or problematic maturational process, trained psychologists assessed the psychomotor development of the children with the Bayley Scales of Infant Development, Second Edition (BSID-II; Bayley, 1993). This widespread tool has received good validity and reliability and allows discrimination between high-risk and low-risk infants (Coleman & Hildebrandt, 2003; Murray & Cooper, 1997a).

Statistics

We used linear regression analyses to test our first hypothesis (the regression analyses described in the following section were performed with PASW Statistics 18). We entered MADRS scores as the predictor (independent variable) and SCL item scores for both mothers and fathers as dependent variables. For the second hypothesis, we used two methods to investigate the moderating role of the family alliance in our sample. First, we computed the same regression analyses described earlier, separately for HC and LC families, to determine whether the MADRS continuous score can differently predict parent ratings of child symptoms in these two groups, which would suggest a moderating effect of the family alliance. Second, we computed hierarchical multiple regression analyses to test the potential moderating effects more precisely (Aiken & West, 2001; Baron & Kenny, 1986; Frazier, Tix, & Barron, 2004). This method consisted of entering the following in a two-step model: (a) the MADRS scores (this variable had been previously centered to avoid multicollinearity problems) and family alliance (dummy coded as 0 = LC and 1 = HC) as predictors in the first step; and (b) a MADRS Scores × Family Alliance variable (namely, the interaction term) in the second step. This analysis was conducted to confirm whether this model can predict the symptoms reported by both parents. Finally, ModGraph, an Excel-based tool, was used to graph our moderating effects and compute the simple slope analyses.

The sample size was reduced from 65 families to 57 due to four missing questionnaires; the removal of one outlier, the data of which were above 3 standard deviations on the MADRS score and the father’s report of the child’s symptoms; and the exclusion of three children whose scores on the motor scale of the BSID-II were less than 77.

Results

Descriptive Statistics

Maternal Depression

As expected in a normative sample, the mean of maternal MADRS scores was low (M = 4.81, SD = 4.79), 15.8% of women (n = 9) having a score exceeding the “mild depression” cutoff (9 points).

Family Interactions

Of the participating families, 64.9% (n = 37) were coded as HC families and 35.1% (n = 20) as LC families. To test the links between the level of maternal depression and the quality of family interactions, we conducted a t test on the mean levels of maternal depression in HC and LC family groups. Even though mean depression was slightly higher in LC than in HC families (M = 6.10 and M = 4.11, respectively), the results of the t test were not significant, t(55) = 1.52, p = .14 (two-tailed).

Fathers’ and Mothers’ Reports of Child Symptoms

Fathers reported different symptoms in their child than mothers did (see details in Table 1). Moreover, fathers rated most of the reported symptoms higher than mothers did: difficulties falling asleep, waking up during the night, sleeping problems, eating problems, opposition, agitation, attention problems, and separation difficulties. However, only one of these differences turned out to be significant: Fathers reported more sleeping problems in their child. Mothers reported more anger crises than fathers did but this difference was not significant.

Table 1.

Detailed Descriptive Statistics for Symptom Check List (SCL) for Mothers and Fathers (n = 57) and Item-by-Item Differences.

	Mothers		Fathers		t test
SCL items	M	SD	M	SD	t	df
1. Difficulties falling asleep	2.40	1.00	2.52	0.96	−0.89	56
2. Waking up during night	2.74	1.34	2.81	1.39	−0.41	56
3. Sleeping problems	1.47	0.83	1.79	1.13	−2.34*	56
5. Eating problems	1.48	0.87	1.53	0.87	−0.28	56
6. Anger crises	2.02	0.95	1.80	0.72	1.42	56
7. Opposition/negativity	1.53	0.67	1.58	0.63	−0.49	56
8. Agitation	2.17	0.83	2.36	0.89	−1.61	56
9. Attention difficulties	2.09	0.84	2.25	0.92	−1.06	56
10 Separation difficulties	1.92	0.93	1.91	0.95	0.11	56

p < .05. item 4 (eating behaviors) was excluded, because it was rated on a bipolar scale from 1 (does not eat enough) to 5 (eats too much).

Maternal Depression as a Predictor of Child Outcomes

First, we computed t tests to compare the mean scores for each child symptom rated with the SCL for mothers above and below the MADRS cutoff (9 points). Mean differences for eight items were not significant. Results showed only one significant difference: Depressed mothers reported more separation difficulties in their baby than did nondepressed mothers, t(55) = −3.139, p < .01 (two-tailed).

To test the relationships between maternal depression and child symptoms, we next conducted linear regression analyses, with MADRS total scores used as a predictor for SCL item ratings of both parents. The level of depression did not appear to be a strong predictor of child symptoms, as rated by both parents. The regression analyses were mostly not significant and globally confirmed the results of the t tests. However, MADRS scores significantly predicted the maternal rating of separation difficulties (Item 10; β = .49, p < .001). Maternal depression accounted for 23% of the variance of these ratings, R² = .23, F(1, 55) = 17.74, p < .001.

Family Alliance as a Moderator

First, regression analyses were computed for each of the LC and HC families, with maternal depression (MADRS scores) as the predictor and the symptoms of the child as dependent variables (SCL, reported by mothers and by fathers).

In LC families, maternal depression did not predict any of the symptoms reported by both parents. On the other hand, in HC families, maternal depression was found to be a strong predictor of four types of symptoms reported by the mothers: difficulties falling asleep, β = .42, R² = .15, F(1, 35) = 7.27, p < .05; eating problems, β = .39, R² = .12, F(1, 35) = 6.11, p < .05; attention problems, β = .40, R² = .13, F(1, 35) = 6.51, p < .05; and separation difficulties, β = .58, R² = .32, F(1, 35) = 17.89, p < .001. These results suggest a moderation effect of the alliance that needed to be tested by the multiple hierarchical regression method.

The results of hierarchical regression analyses partially confirmed the results (Table 2). The interaction term Depression × Alliance significantly predicted the report of difficulties falling asleep and attention problems by mothers. For difficulties falling asleep, family alliance and depression did not separately explain a significant part of the variance. On the other hand, the interaction between these two variables entered in a second step produced a significant R² change, explaining an extra 8% of variance (more than the 3% explained by Step 1). The results for the attention problems taken as outcome variables were similar, as both predictor (MADRS score centered) and moderator (family alliance) entered separately at Step 1 did not account for a significant proportion of the outcome variance (R² = .03, p = .35). Entering the interaction term at Step 2 produced a significant R² change, accounting for 9% more of the variance in mothers’ scores on this item. Such moderating effects are illustrated in Figure 1.

Table 2.

Multiple Hierarchical Regression Analyses Testing the Moderating Effect of Family Alliance (LC vs. HC) on the Relationship Between Depression (MADRS) and Child Symptoms Reported by Mothers (SCL).

Step and variable	B	SE B	95% CI	R ²
	Predicting maternal report of “difficulties falling asleep”
Step 1
Alliance (dummy: 0 = LC, 1 = HC)	−.17	.08	−.73, .40	.03
Depression (centered)	.30	.03	−.03, .09	.03
Step 2
Alliance × Depression	.12	.06	.01, .23	.08*
	Predicting maternal report of “attention difficulties”
Step 1
Alliance (dummy: 0 = LC, 1 = HC)	.30	.24	−.18, .78	.04
Depression (centered)	.02	.02	−.02, .07	.04
Step 2
Alliance × Depression	.12	.05	.01, .20	.09*

Note. The results are displayed according to the recommendations of Frazier et al. (2004) and Aiken and West (1991). Only unstandardized coefficients are listed.

CI = confidence interval; HC = high coordination families; LC = low coordination families; MADRS = Montgomery and Asberg Depression Rating Scale; SCL = Symptom Check List.

p < .05.

Figure 1.

Moderating effect of family alliance on the link between maternal depression scores (Montgomery and Asberg Depression Rating Scale [MADRS]) and difficulties falling asleep (top) and attentional difficulties (bottom) in children as reported by mothers (Symptom Check List [SCL]).

Figure 1 is a graphic representation of the links between MADRS and the maternal report of difficulties falling asleep and attention problems for HC and LC families. The simple slope analyses revealed that, in HC families, there are small but significant correlations between the depressive level of mothers and their report of their child’s difficulties in falling asleep (r = .09, p < .01) and attention difficulties (r = .08, p < .05). In contrast, these relationships were not significant in LC families.

Discussion

Maternal Depression as a Predictor of Child Outcomes

The results show that the more a mother is depressed, the more she describes her 3-month-old child as having separation difficulties. This finding is in line with other studies showing that when the child is 18 months old, separation difficulties are more frequently reported by mothers who develop PPD (Murray & Cooper, 1997b). This suggests that separation difficulties may reflect the depressed mothers’ own difficulties in establishing early relationships with their baby, which may take root in the very early months of life and thus be already present at 3 months.

However, beyond this isolated significant result, maternal depression seems to be a relatively weak predictor of early child symptoms in this community sample, in contrast to what has been demonstrated in studies with clinical samples. Authors such as Sameroff (1998) underlined that the effects of parental psychopathology on the child and the family were most likely to be stronger in families where risk factors have accumulated. Studies about maternal PPD in low-risk families confirmed these observations, identifying no effect or very small effect sizes of PPD on the child (e.g., see Weinberg et al., 2008). In the present study, we found a rate of depressed mothers (above the MADRS cutoff) in line with the prevalence rates of PPD reported in the literature, with most depressed mothers situated in the mild depression range, as the highest score registered was 22 points (out of 60). These low depression scores, as well as the sociodemographic data (most of our families were of upper-middle socioeconomic status), reflected the low-risk characteristics of the families included in this study. In that sense, our results are quite consistent with those of other studies. Other variables may thus have played a moderating role on the effect of PPD on the children in our sample, such as the aforementioned high socioeconomic status, marital status (no single parents), or level of social support (most of the families in our study reported that they could find an important support in their close social environment). The precise impact of maternal depression on the child in low-risk families is still unclear and might be difficult to investigate because of the low effect sizes that can be expected. Attempting to improve our knowledge about it is worth the challenge, however, as it is of key importance in enriching our understanding of these situations. Conducting studies with larger samples may help us detect these low effect sizes’ relationships between the variables under study.

The Moderating Effect of Family Alliance

Our hypothesis concerning the moderating effect of family alliance on the link between PPD and child developmental outcomes was partially confirmed but in an unexpected way. When we compared the PPD as a predictor of child symptoms in HC and LC families, we found the first confirmation of the moderating effect of the family alliance. Indeed, maternal depression level seems to be a significant predictor of four types of child symptoms, as reported by mothers in HC families. In contrast, no link was found between PPD and child symptoms in LC families. This result was partially confirmed by hierarchical regression analyses, as the cross-product interaction between the level of maternal depression and family alliance was a better predictor of child outcomes than both variables alone for two symptoms, explaining a significant proportion of the additional variance. At first, we hypothesized that we may find more links between maternal depression and child symptoms in LC than in HC families, considering that positive family relationships may be a possible buffer against the impact of maternal depression on the child. In fact, our results supported the opposite perspective because the links between PPD and child symptoms were stronger in families with more positive relationships.

These results, albeit surprising, turned out to be in line with the results of a previous study about the role of family alliance. In a recent article, Favez et al. (2011) hypothesized a positive correlation between family alliance and marital satisfaction, as it has been established by other studies in the field, as well as in clinical descriptions, that marital and family-level functioning seems to be highly correlated (Hetherington, 1992; Katz & Low, 2004). Surprisingly, all the correlations turned out to be negative: The better the family alliance, the less that marital satisfaction was reported. The authors explained this unexpected result by arguing that parents in more functional family contexts might be more able to develop reflexive thinking and thus to be more critical of their marital relationship, whereas parents facing difficulties within the family might trigger defense mechanisms, such as denial, leading to a bias in the assessment of their marital relationship in a falsely positive way.

In line with this finding, we assume that a sense of security and emotional kindliness may be brought to mothers in HC families by their partner’s support and by positive family relationships in general, which may help them to face problems in the family, such as the emergence of symptoms in early childhood. In contrast, relational difficulties may lead the mothers in LC families to avoid negative assessments or answers in the questionnaires about their child (and possibly about themselves and marital or family relationships) because of a high activation of defense mechanisms resulting in high social desirability. This phenomenon, also called defensive or fake good responding (Bagby, Buis, & Nicholson, 1995), has been described as a common bias in psychological measurement and has been more specifically addressed in parenting-related measures such as parenting stress (Abidin, 1995). In the present study, the use of denial or other immature-level defense mechanisms may therefore explain the absence of links between PPD and child outcomes in LC families. A complementary methodological approach, such as a questionnaire filled in by another person in the extended family or by any secondary caregiver, will be necessary to verify this finding.

With these arguments in mind, we postulate that the child symptoms reported by the parents are influenced by their perception of the child, suggesting that their report will be biased and distorted by their subjective involvement in the situation. Even though this may be the case, we cannot deny the fact that the development of children in families with positive relationships may in fact be more hampered by maternal depression. Some arguments can be put forward to support this point of view. First, the family alliance model assumes that HC families can reach a high coordination in triadic interactions thanks to stronger emotional and empathetic bonds between the family members: In the context of maternal depression in this type of family, the negative emotions of the depressive mood could invade these bonds and thus be more easily transmitted to the other members than occurs in LC families. This may explain why depressive features may have a stronger impact in children of HC families. In addition, fathers in HC families may also be more at risk to be affected by their partner’s depression. Consequently, it may be more difficult for such a father to act as a buffer against the impact of maternal depression on the child. Second, adopting such systemic points of view may also help change our perspective about the role of family relationships. Indeed, maternal depression might have different and specific impacts from one family to the other: In some families, depression may have a negative impact on the emergence of symptoms in the child (in HC families), whereas in others, the impact may operate on the family level, deteriorating the quality of the relationships (LC families). This implies that family relationships are not the moderator but rather are dependent variables that could also be influenced by the occurrence of maternal depression. This last perspective supports research showing that maternal psychopathology has a strong negative link to family functioning (Dickstein et al., 1998). Such questions should be examined and specifically addressed in further studies.

Implications of the Findings

Our results show that the impact of PPD on child development may also depend on maternal ability to perceive the child’s difficulties and to “ring the alarm” when needed. From a long-term perspective, this ability may lead mothers to seek help sooner for their child (and possibly for themselves), which may reduce the damage of maternal depression on the child. In that sense, we can surmise that the sooner the problems are detected and taken care of, the less impact they will exert on the child.

These results reinforce the need to develop secondary prevention programs on a large scale to increase the knowledge of the general population about the disorder and its adverse effects on the child. In our sample, none of the 15% of mothers who had mild depression were seeking help. Thus, early prevention programs set up in the prenatal or immediate postnatal periods are needed (Dennis & Chung-Lee, 2006). This idea was reinforced in our results by the fact that, at least in some families, mild PPDs could already have an impact on children at 3 months. This finding is of key importance, as it underlines how fast psychopathological disorders may develop in this sensitive period. The baby’s well-being may rapidly be jeopardized by the short-term effects of PPD, which can affect several aspects of her functioning. Our study also suggests that the specific problems reported by mothers (such as separation difficulties) or excessive concerns about the baby’s health may be used as a potential indicator of depression in the mother. Social workers or medical staff working with postnatally depressed mothers should be aware that a child may be silently suffering while the parents do not report any problems. Family therapists may pay specific attention to this phenomenon, as our results suggest that this could more likely occur in distressed families, such as the LC families in the present research. These results have to be confirmed in further studies with a sample of high-risk families, which are distressed not only at a relational level (like the LC families in our studies) but also at a psychosocial level.

Our results suggest that there are strong links between maternal well-being, mother–infant relationship, and family-level relationship. Adopting a comprehensive perspective of these levels of family functioning may help the clinicians to improve the treatments for postnatally depressed mothers, their child, and their family. We think that it could be useful for every clinicians taking care of depressed mothers, even with a low level of symptoms, to look after the possible causes and consequences of the maternal distress within the whole family system.

The goal of this study was to test hypotheses concerning the links between PPD, child development, and family-level variables for families in the general population, as PPD is known to be underdiagnosed and thus undertreated. The findings underline the relatively high rates of depressed mothers and the links between maternal depression and child outcomes in the child’s very first few months, calling for a more precise investigation into the psychosocial context in which depressive disorders occur. We hope that research in the field will also lead to strategies for the prevention of early childhood disorders for children of depressed mothers and to an increase in the sensitivity of people in the health care system and in the general population about PPD and its consequences.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Swiss National Science Foundation (SNSF) Grant 32003B_125493

References

Abidin

R. R.

(1995). Parenting Stress Index (3rd ed.). Odessa, FL: Psychological Assessment Resources.

Aiken

L. S.

West

S. G.

(1991). Multiple regression: Testing and interpreting interactions. Thousand Oaks, CA: Sage.

American Psychological Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed., text revision). Washington, DC: Author.

Bagby

R. M.

Buis

Nicholson

R. A.

(1995). Relative effectiveness of the standard validity scales in detecting fake-bad and fake-good responding: Replication and extension. Psychological Assessment, 7, 84-92. doi:10.1037/1040-3590.7.1.84

Baron

R. M.

Kenny

D. A.

(1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1182.

Bayley

(1993). Bayley Scales of Infant Development (2nd ed.). San Antonio, TX: Psychological Corporation.

Beebe

Jaffe

Buck

Chen

Cohen

Feldstein

Andrews

(2008). Six-week postpartum maternal depressive symptoms and 4-month mother–infant self-and interactive contingency. Infant Mental Health Journal, 29, 442-471. doi:10.1002/imhj.20191

Bost

K. K.

Cox

M. J.

Burchinal

M. R.

Payne

(2002). Structural and supportive changes in couples’ family and friendship networks across the transition to parenthood. Journal of Marriage and Family, 64, 517-531. doi:10.1111/j.1741-3737.2002.00517.x

Carro

M. G.

Grant

K. E.

Gotlib

I. H.

Compas

B. E.

(1993). Postpartum depression and child development: An investigation of mothers and fathers as sources of risk and resilience. Development and Psychopathology, 5, 567-579. doi:10.1017/S0954579400006167

10.

Cicchetti

Rogosch

F. A.

Toth

S. L.

(1998). Maternal depressive disorder and contextual risk: Contributions to the development of attachment insecurity and behavior problems in toddlerhood. Development and Psychopathology, 10, 283-300. doi:10.1017/S0954579498001618

11.

Cogill

Caplan

Alexandra

Robson

Kumar

(1986). Impact of maternal postnatal depression on cognitive development of young children. British Medical Journal, 292, 1165-1167. doi:10.1136/bmj.292.6529.1165

12.

Coleman

P. K.

Hildebrandt

(2003). Maternal self-efficacy beliefs, competence in parenting, and toddlers’ behavior and developmental status. Infant Mental Health Journal, 24, 126-148. doi:10.1002/imhj.10048

13.

Corboz-Warnery

Fivaz-Depeursinge

Gertsch Bettens

Favez

(1993). Systemic analysis of father-mother-baby interactions: The Lausanne triadic play. Infant Mental Health Journal, 14, 298-316.

14.

Dennis

C.-L.

Chung-Lee

(2006). Postpartum depression help-seeking barriers and maternal treatment preferences: A qualitative systematic review. Birth, 33, 323-331. doi:10.1111/j.1523-536X.2006.00130.x

15.

Dix

Cheng

Day

W. H.

(2009). Connecting with parents: Mothers’ depressive symptoms and responsive behaviors in the regulation of social contact by one- and young two-year-olds. Social Development, 18, 24-50. doi:10.1111/j.1467-9507.2008.00488.x

16.

Dickstein

Seifer

Hayden

L. C.

Schiller

Sameroff

A. J.

Keitner

Magee

K. D.

(1998). Levels of family assessment: II. Impact of maternal psychopathology on family functioning. Journal of Family Psychology, 12, 23-40. doi:10.1037//0893-3200.12.1.23

17.

Edhborg

Lundh

Seimyr

Widström

A. M.

(2003). The parent-child relationship in the context of maternal depressive mood. Archives of Women’s Mental Health, 6, 211-216. doi:10.1007/s00737-003-0020-x

18.

Favez

Frascarolo

Carneiro

Montfort

Corboz-Warnery

Fivaz-Depeursinge

(2006). The development of the family alliance from pregnancy to toddlerhood and children outcomes at 18 months. Infant and Child Development, 15, 59-73. doi:10.1002/icd.430

19.

Favez

Frascarolo

Fivaz-Depeursinge

(2006). Family alliance stability and change from pregnancy to toddlerhood and marital correlates. Swiss Journal of Psychology, 65, 213-220. doi:10.1024/1421-0185.65.4.213

20.

Favez

Lavanchy Scaiola

Tissot

Darwiche

Frascarolo

(2011). The Family Alliance Assessment Scales: Steps toward validity and reliability of an observational assessment tool for early family interactions. Journal of Child and Family Studies, 20, 23-37. doi:10.1007/s10826-010-9374-7

21.

Field

(2010). Postpartum depression effects on early interactions, parenting, and safety practices: A review. Infant Behavior and Development, 33, 1-6. doi:10.1016/j.infbeh.2009.10.005

22.

Field

Hernandez-Reif

Diego

(2006). Intrusive and withdrawn depressed mothers and their infants. Developmental Review, 26, 15-30. doi:10.1016/j.dr.2005.04.001

23.

Fivaz-Depeursinge

Corboz-Warnery

(1999). The primary triangle: A developmental systems view of mothers, fathers, and infants. New York, NY: Basic Books.

24.

Fivaz-Depeursinge

Frascarolo

Lopes

Dimitrova

Favez

(2007). Parents–child role reversal in trilogue play: Case studies of trajectories from pregnancy to toddlerhood. Attachment & Human Development, 9, 17-31. doi:10.1080/14616730601151425

25.

Frascarolo

Favez

Carneiro

Fivaz-Depeursinge

(2004). Hierarchy of interactive functions in father-mother-baby three-way games. Infant and Child Development, 13, 301-322. doi:10.1002/icd.361

26.

Frazier

P. A.

Tix

A. P.

Barron

K. E.

(2004). Testing moderator and mediator effects in counseling psychology research. Journal of Counseling Psychology, 51, 115-134. doi:10.1037/0022-0167.51.1.115

27.

Goodman

S. H.

Brogan

Lynch

M. E.

Fielding

(1993). Social and emotional competence in children of depressed mothers. Child Development, 64, 516-531. doi:10.1111/j.1467-8624.1993.tb02925.x

28.

Goodman

S. H.

Brumley

H. E.

(1990). Schizophrenic and depressed mothers: Relational deficits in parenting. Developmental Psychology, 26, 31-39.

29.

Gotlib

I. H.

Beach

S. R. H.

(1995). A marital/family discord model of depression: Implications for therapeutic intervention. In Jacobson

N. S.

Gurman

A. S.

(Eds.), Clinical handbook of couple therapy (pp. 411-436). New York, NY: Guilford Press.

30.

Hay

D. F.

(1997). Postpartum depression and cognitive development. In Murray

Cooper

P. J.

(Eds.), Postpartum depression and child development (pp. 85-110). New York, NY: Guilford Press.

31.

Hay

D. F.

Kumar

(1995). Interpreting the effects of mothers’ postnatal depression on children’s intelligence: A critique and re-analysis. Child Psychiatry and Human Development, 25, 165-181. doi:10.1007/BF02251301

32.

Hetherington

E. M.

(1992). Coping with marital transitions: A family systems perspective. Monographs of the Society for Research in Child Development, 57, 1-14. doi:10.1111/j.1540-5834.1992.tb00300.x

33.

Hollingshead

A. B.

(1975). Four factor index of social status (Unpublished manuscript). Yale University, New Haven, CT.

34.

Katz

L. F.

Low

S. M.

(2004). Marital violence, co-parenting, and family-level processes in relation to children’s adjustment. Journal of Family Psychology, 18, 372-382. doi:10.1037/0893-3200.18.2.372

35.

Knudson-Martin

Silverstein

(2009). Suffering in silence: A qualitative meta-data-analysis of postpartum depression. Journal of Marital and Family Therapy, 35, 145-158. doi:10.1111/j.1752-0606.2009.00112.x

36.

Leckman-Westin

Cohen

P. R.

Stueve

(2009). Maternal depression and mother–child interaction patterns: Association with toddler problems and continuity of effects to late childhood. Journal of Child Psychology and Psychiatry, 50, 1176-1184. doi:10.1111/j.1469-7610.2009.02083.x

37.

Lemperière

Lépine

Rouillon

Hardy

Adès

Luaute

Ferrand

(1984). Comparaison de différents instruments d’évaluation de la dépression à l’occasion d’une étude sur l’Athymil 30 mg [Comparison of different tools designed to assess depression in a study about Athymil 30 mg]. Annale Medico-Psychologique, 142, 1206-1212.

38.

Lyons-Ruth

Lyubchik

Wolfe

Bronfman

(2002). Parental depression and child attachment: Hostile and helpless profiles of parent and child behavior among families at risk. In Goodman

Gotlib

(Eds.), Children of depressed parents (pp. 89-120). Washington, DC: American Psychological Association. doi:10.1037/10449-004

39.

Manzano

Righetti-Veltema

Conne Perreard

(1997). Le syndrome de dépression du pré-partum. Résultat d’une recherche sur les signes précurseurs de la dépression du post-partum [Prepartum depression syndrome. The results of a study about the predictors of postpartum depression]. La Psychiatrie de l’Enfant, 40, 533-552.

40.

McElwain

N. L.

Volling

B. L.

(1999). Depressed mood and marital conflict: Relations to maternal and paternal intrusiveness with one-year-old infants. Journal of Applied Developmental Psychology, 20, 63-83. doi:10.1016/S0193-3973(99)80004-5

41.

McGrath

J. M.

Records

Rice

(2008). Maternal depression and infant temperament characteristics. Infant Behavior and Development, 31, 71-80. doi:10.1016/j.infbeh.2007.07.001

42.

Minuchin

(1974). Families and family therapy. Boston, MA: Harvard University Press.

43.

Montgomery

S. A.

Asberg

(1979). A new depression scale designed to be sensitive to change. British Journal of Psychiatry, 134, 382-389. doi:10.1192/bjp.134.4.382

44.

Müller

M. J.

Szegedi

Wetzel

Benkert

(2000). Moderate and severe depression: Gradations for the Montgomery-Åsberg Depression Rating Scale. Journal of Affective Disorders, 60, 137-140. doi:10.1016/S0165-0327(99)00162-7

45.

Murray

(1992). The impact of postnatal depression on infant development. Journal of Child Psychology and Psychiatry, 33, 543-561. doi:10.1111/j.1469-7610.1992.tb00890.x

46.

Murray

Cooper

P. J.

(1997a). The role of infant and maternal factors in postpartum depression, mother-infant interactions, and infant outcome. In Murray

Cooper

P. J.

(Eds.), Postpartum depression and child development (pp. 111-135). New York, NY: Guilford Press.

47.

Murray

Cooper

(1997b). Effects of postnatal depression on infant development. Archives of Disease in Childhood, 77, 99-101.

48.

Nurnberg

Blehar

Kaufmann

York-Cooler

Simpson

Harkavy-Friedman

Reich

(1994). Diagnostic interview for genetic studies: Rationale, unique features, and training. Archives of General Psychiatry, 51, 849-859.

49.

O’Hara

M. W.

(1994). Postpartum depression: Causes and consequences. New York, NY: Springer Verlag.

50.

O’Hara

M. W.

(1997). The nature of postpartum depressive disorders. In Murray

Cooper

P. J.

(Eds.), Postpartum depression and child development (pp. 3-31). New York, NY: Guilford Press.

51.

Papousek

(1997). Fragile aspects of early social integration. In Murray

Cooper

P. J.

(Eds.), Postpartum depression and child development (pp. 35-53). New York, NY: Guilford Press.

52.

Paulson

J. F.

Dauber

S. E.

Leiferman

J. A.

(2011). Parental depression, relationship quality, and nonresident father involvement with their infants. Journal of Family Issues, 32, 528-549.

53.

Pitt

(1968). “Atypical” depression following childbirth. British Journal of Psychiatry, 114, 1325-1335. doi:10.1192/bjp.114.516.1325

54.

Preisig

Fenton

B. T.

Matthey

M. L.

Berney

Ferrero

(1999). Diagnostic interview for genetic studies (DIGS): Inter-rater and test-retest reliability of the French version. European Archives of Psychiatry and Clinical Neuroscience, 249, 174-179. doi:10.1007/s004060050084

55.

Robert-Tissot

Rusconi-Serpa

Bachman

Besson

Cramer

Knauer

Stern

D. N.

(1989). Le questionnaire “Symptom Check-List”: Evaluation des troubles psychofonctionnels de la petite enfance [The “Symptom Check-List” questionnaire: Assessing psycho-functional disorders in early childhood]. In Lebovici

Mazet

(Eds.), L’évaluation des interactions précoces entre le bébé et ses partenaires [Assessing the early interactions between the baby and her partners] (pp. 179-215). Paris, France: Eshel.

56.

Robert-Tissot

Rusconi-Serpa

De Muralt

Stern

Cramer

(1991). Traitement des troubles fonctionnels dans la petite enfance [The treatment of psycho-functional disorders in early childhood]. Cahiers Psychiatriques Genevois, 11, 109-123.

57.

Ross

Evans

Sellers

Romach

(2003). Measurement issues in postpartum depression. Part 1: Anxiety as a feature of postpartum depression. Archives of Women’s Mental Health, 6, 51-57. doi:10.1007/s00737-002-0155-1

58.

Sameroff

A. J.

(1998). Environmental risk factors in infancy. Pediatrics, 102(Suppl. E1), 1287-1292.

59.

Sheeder

Kabir

Stafford

(2009). Screening for postpartum depression at well-child visits: Is once enough during the first 6 months of life? Pediatrics, 123, 982-988. doi:10.1542/peds.2008-1160

60.

Teichman

(1990). Interpersonal view of depression: Review and integration. Journal of Family Psychology, 3, 349-367. doi:10.1037//0893-3200.3.4.349

61.

Tharner

Luijk

M. P. C. M.

van Ijzendoorn

M. H.

Bakermans-Kranenburg

M. J.

Jaddoe

V. W. V.

Hofman

Tiemeier

(2011). Maternal lifetime history of depression and depressive symptoms in the prenatal and early postnatal period do not predict infant–mother attachment quality in a large, population-based Dutch cohort study. Attachment & Human Development, 14, 63-81. doi:10.1080/14616734.2012.636659

62.

Tissot

(2009). Le Family Alliance Assessment Scale (FAAS). Premiers pas vers la validation d’un système de codage des interactions familiales [The Family Alliance Assessment Scale (FAAS). First steps toward the validation of a coding system of family interactions] (Unpublished master’s thesis). University of Lausanne, Switzerland.

63.

Tronick

E. Z.

Gianino

A. F.

(1986). The transmission of maternal disturbance to the infant. New Directions for Child and Adolescent Development, 34, 31-47. doi:10.1002/cd.23219863403

64.

Tronick

E. Z.

Weinberg

M. K.

(1997). Depressed mothers and infants: Failure to form dyadic states of consciousness. In Murray

Cooper

P. J.

(Eds.), Postpartum depression and child development (pp. 54-81). New York, NY: Guilford Press.

65.

Vliegen

Luyten

Biringen

(2009). A multimethod perspective on emotional availability in the postpartum period. Parenting, 9, 228-243. doi:10.1080/15295190902844514

66.

Weinberg

M. K.

Olson

Beeghly

Tronick

E. Z.

(2008). Effects of maternal depression and panic disorder on mother-infant interactive behavior in the face-to-face still-face paradigm. Infant Mental Health Journal, 29, 472-491. doi:10.1002/imhj.20193

67.

Weinberg

M. K.

Tronick

E. Z.

Beeghly

Olson

K. L.

Kernan

Riley

J. M.

(2001). Subsyndromal depressive symptoms and major depression in postpartum women. American Journal of Orthopsychiatry, 71, 87-97. doi:10.1037/0002-9432.71.1.87

68.

World Health Organization. (1992). The ICD-10 classification of mental and behavioural disorders: Clinical descriptions and diagnostic guidelines. Geneva, Switzerland: Author.