Infant Behavior & Development 40 (2015) 183–192
Contents lists available at ScienceDirect
Infant Behavior and Development
Full length article
Prenatal depression and young low-income mothers’ perception of their children from pregnancy through early childhood Helen Y. Lee a,∗ , Sydney L. Hans b a b
Boston University, Boston MA, United States University of Chicago, Chicago, IL, United States
a r t i c l e
i n f o
Article history: Received 27 August 2014 Received in revised form 25 February 2015 Accepted 2 June 2015 Available online 10 July 2015 Keywords: Early childhood Longitudinal data Mental representation Prenatal depression Perception Pregnancy
a b s t r a c t This study investigates the role of prenatal depression on young mothers’ perception of difficulty in child behavior and parent–child interaction from pregnancy through the first two years of child development. 248 low-income, African American women aged between 13 and 21 years reported on their perceptions of child behavior, parent–child interaction, and on depressive symptoms at the third trimester of pregnancy and at 4-, 12- and 24months postpartum. During pregnancy, a high percentage (47%) of the young mothers scored above the clinical level of prenatal depressive symptoms. These mothers anticipated and perceived significantly more child difficulty and parent–child interaction difficulty than did non-depressed mothers during pregnancy and over the first two years of the child’s development. Moreover, prenatal depression uniquely predicted negative maternal perception throughout the early years of child development even after adjusting for postpartum depression history. The enduring association between prenatal depression and a mother’s perception of her child is discussed with respect to the importance of pregnancy in mothers’ developing mental schema about their children and the emerging parent–child relationship. © 2015 Elsevier Inc. All rights reserved.
1. Introduction Depression during pregnancy is prevalent among women in the United States, affecting up to 38% of all pregnant women (Anderson, Sundstrom-Poromaa, Wulff, Astrom, & Bixo, 2003; Records & Rice, 2007) and 60% of at-risk mothers, including low-income teenage mothers (Earls & The Committee on Psychosocial Aspects of Child & Family Health, 2010; Goyal, Gay, & Lee, 2010). In recent years, the adverse effects of prenatal depression on maternal health and child development have received increasing attention in the literature, shedding light on its association with poor early mother–child interaction and child outcomes (for review see Field, Diego, & Hernandez-Reif, 2006) and its association with postpartum depression (Larsson, Sydsjo, & Josefsson, 2004; Milgrom et al., 2008). Much of the literature on prenatal depression to date has focused on the immediate and direct influence of maternal prenatal depression on the child via biochemical pathways in the womb environment (e.g., Lester, Conradt, & Marsit, 2013). However, yet another major way in which prenatal depression may impact the child after birth, which has not been
∗ Corresponding author. E-mail address: [email protected] (H.Y. Lee). http://dx.doi.org/10.1016/j.infbeh.2015.06.008 0163-6383/© 2015 Elsevier Inc. All rights reserved.
184
H.Y. Lee, S.L. Hans / Infant Behavior & Development 40 (2015) 183–192
well-explored, is through maternal cognition. The negative cognitions that depressed women establish about their unborn child during pregnancy may be patterns of thought that are the basis for future views of their children and their parent–child relationship, as well as their parenting behavior. A growing theoretical and empirical literature has highlighted the importance of parents’ representations of their children as a central feature of the parent–child relationship (e.g., Bugental & Happaney, 2002; Sigel, McGillicuddy-DeLisi, & Goodnow, 2014; Stern, 1989). Such representations are complex cognitive schemas of the child that are linked to emotions and to representations of self and others in the parent’s experience. Maternal representations of the child have been shown to be linked to maternal behavior, to infant attachment, and to infant development more generally (e.g., Broussard & Cassidy, 2010; Nover, Shore, Timberlake, & Greenspan, 2009; Zeanah & Barton, 1989). In particular, negative-valenced representations of the child, such as children being difficult, have been linked to developmental risk (Broussard & Cassidy, 2010). Mothers’ representations of the child begin before the birth of the child, and pregnancy is a crucial time when women anticipate and actively form ideas about their unborn children and about the impending parent–child relationship that forms the basis for enduring beliefs about the child (Ammaniti, Tambelli, & Odorisio, 2013; Lashley, Hans, & Henson, 2009; Nelson, 2003; Pajulo, Savonlahti, Sourander, Helenius, & Piha, 2001; Theran, Levendosky, Bogat, & Huth-Bocks, 2005; Zeanah, Keener, Stewart, & Anders, 1985). Depressive symptoms are closely linked to the ways mothers perceive and think about their children and their relationships with the children (Goodman & Gotlib, 1999; Joiner & Coyne, 1999). Depressed individuals show generally negative cognitive biases when processing social and affective information from interpersonal contexts (see review Bistricky, Ingram, & Atchley, 2011). They tend to selectively attend to and recollect negative or incongruent materials (Beck & Perkins, 2001; Gotlib, Krasnoperova, Yue, & Joormann, 2004), ruminate on negative information, and show more difficulty disengaging attention from negatively valenced materials than do non-depressed individuals (see review Gotlib & Joormann, 2010). As parents, women with depressive symptoms perceive more difficulty and problems with diverse aspects of their children and parent–child interactions across infancy to childhood (Ingersoll & Eist, 1998; Lovejoy, Graczyk, O’Hare, & Neuman, 2000; Weissman & Paykel, 1974). For example, mothers with depression perceive more difficulty in their newborns’ basic regulatory behavior, such as eating and bowel movements (Fulton, Mastergeoge, Steele, & Hansen, 2012), in temperamental characteristics of infants related to irritability and fussiness (McGrath, Records, & Rice, 2008), and in conduct problems during late childhood and adolescence (Conrad & Hammen, 1989; Galler, Harrison, Ramsey, Butler, & Forde, 2004; Teti, Gelfand, & Pompa, 1990; Whiffen & Gotlib, 1989). Despite the high prevalence of prenatal depression and the saliency of emerging parental cognition about the child and the parent–child relationship during pregnancy, the implication of depression during pregnancy on mothers’ immediate and longer-term perceptions of their children has been little examined. Some data suggest that prenatally depressed mothers may have a less strong emotional connection toward their fetus than other women (Lindgren, 2001; McFarland et al., 2011; Scharfe, 2007), but we know little about how prenatal depression affects women’s perceptions of their unborn children or predictions for what their children will be like after birth. Several studies suggest that prenatal maternal depression is associated longitudinally with mothers’ perceptions after birth of their infants as being difficult to care for. McGrath et al. (2008) found that prenatal depression was associated with reports of difficult temperament at two and six months after birth, although they did not control for maternal postnatal depression. Huot, Brennan, Stowe, Plotshky and Walker (2004) showed that prenatal maternal depression and not postnatal maternal depression related to mothers’ ratings of infant negative affect expression at six months of age. Similarly, Della Vedova (2014) found that women who are depressed during pregnancy rated their infants’ temperament as more difficult at three months after birth. A clinical sample of women with prenatal major depressive disorder were also relatively unlikely at three months to have balanced representations of their infants—characterized by positive, rich, and full descriptions of their infants (Wood, Hargreaves, & Marks, 2004). There are two main aims for the present study. First, the study aims to describe prenatally depressed mothers’ perceptions of child difficulty from the prenatal period through the first two years of child development, including the first hours after birth, and to compare their perceptions to those of non-depressed mothers. Little is currently known about the trajectory of changes in perception between depressed and non-depressed mothers from pregnancy to postpartum, especially in the earliest days of the neonatal period when mothers have not had an opportunity to revise their ideas about the child based on interaction with the child. Second, the study aims to examine the contribution of prenatal depression on mothers’ perception of their children over time by taking later depressive symptoms into account. Although the independent effect of prenatal depression after adjusting for postpartum depression has been previously found with respect to later child outcomes (Deave, Heron, Evans, & Emond, 2008; Pearson et al., 2012; Hay, Pawlby, Waters, Perra, & Sharp, 2010), it has not been examined with respect to maternal perceptions of the child. Moreover, considering that previous literature has suggested that the history of depression is a major predictor of mothers’ postpartum depression (for review Lancaster et al., 2010), the present study was designed to adjust for the prior postpartum depression status in addition to mothers’ concurrent depression status. 2. Methods 2.1. Participants 248 young women were recruited from two prenatal care clinics affiliated with a large Midwestern university hospital system as a part of a larger intervention study. The women were eligible to participate in the study if they were under 22
H.Y. Lee, S.L. Hans / Infant Behavior & Development 40 (2015) 183–192
185
Table 1 Demographic characteristics of participants. Characteristics
M
SD
Mother age in years (N = 248) Highest education level completed in years (N = 247)
17.6 10.7 n
1.7 1.6 %
30 22 17 142 18 2 1
12.9 9.5 7.3 61.2 7.8 0.9 0.4
218 29 1
87.5 12.1 0.4
128 115
52.7 47.3
174 71
71.0 29.0
Relationship with father of baby (N = 232) No relationship Some level of friendship Good friends Boyfriend Fiancé Married Other Participant’s other children (N = 248) 0 1 2 Gender of baby (N = 243) Boy Girl Neonatal Intensive Care Unit (NICU) experience (N = 245) No Yes
years of age, less than 34 weeks gestation, not planning to move from the area, and willing to participate in an intervention study offering health and childbirth education and support. All of the young mothers recruited from the clinics were AfricanAmerican. At enrollment, the mothers’ ages ranged between 13 and 21 years, with a mean age of 17.6 (SD = 1.7) years. Most mothers received Medicaid (94% or N = 233). Mothers had completed an average of 10.7 years of school (SD = 1.6), and 37.9% (N = 94) had completed high school. 13% (N = 30) of the women reported having no relationship with the father of their baby during pregnancy, 86% (N = 201) reported having an ongoing friendship or romantic relationship with the father of their baby, and 1% (N = 2) reported being married. 88% of the women (N = 203) were first-time mothers. The mean weight of the children was 6.80 pounds (SD = 1.32) at birth. 29% of the children (N = 71) were housed in the hospital Neonatal Intensive Care Unit (NICU) after birth, rather than allowed to room in with their mother because of minor to severe medical problems with respect to the infant or mother. Table 1 shows detailed demographic information. The mothers were interviewed at five time points: pregnancy (mean 84 days before giving birth), 1 day after the birth, and 4-, 12-, and 24-months after the birth. 90.7% (N = 235) of the sample was retained for interviews at 1 day, 89.1% (= 221) at 4 months, 88.3% (n = 219) at twelve months and 79.4% (n = 219) at 12 months and 79.4% ( = 197) at 24 months. Reasons for loss of data at one day were primarily mothers giving birth at non-study hospitals or maternal and infant medical complications that made interviewing difficult. At later waves, data loss was primarily due to inability to locate mothers or to schedule interviews within an appropriate time frame, although 3 infants died as newborns, 1 mother withdrew from study participation, and 3 mothers lost or surrendered custody of their infants. Some mothers who were lost at one wave returned to the sample at later longitudinal assessment points. Analyses were conducted to examine potential differential loss to follow up based on a set of 20 demographic, medical, and psychosocial characteristics measured at enrollment. Retention was related to only one of these baseline variables: mothers who had been working at enrollment were less likely to be participating in the study at 4, 12, and 24 months. Mothers who had high prenatal depressive symptoms were not significantly more likely to be lost to follow up at 4, 12 or 24 months (12.0%, 14.5%, and 23.1%, respectively) than were mothers with low prenatal depressive symptoms (11.5%, 9.2%, 17.6%). 2.2. Procedure Maternal depressive symptoms were assessed with the Center for Epidemiology Studies Depression Scale (CES-D) at each time point, except at 1 day after birth. Maternal perceptions of the infant prenatally and 1 day postpartum were measured by the Neonatal Perception Inventory (NPI). Mothernal perceptions of the older infants were measured when the child was 4-, 12-, and 24-months old by the Parenting Stress Index (PSI). Demographic and medical covariates were assessed prenatally or neonatally. 2.2.1. Measures Center for Epidemiological Studies Depression Scale (CES-D; Radloff, 1977). This self-report 20-item questionnaire surveys mood, somatic complaints, interactions with others, and motor functioning. The response values are 4-point Likert scales, with a range of 0-3, with ‘0 = rarely or none of the time (less than one day)’ to ‘3 = most or all of the time (5-7days).’ The final sum score can be anywhere between 0 and 60, with a higher score indicating greater impairment, and classifies a total score of 16 or higher as a depressive case according to the instrument’s authors. For the purpose of all statistical analysis
186
H.Y. Lee, S.L. Hans / Infant Behavior & Development 40 (2015) 183–192
Table 2 Means and standard deviations for maternal depressive symptom scores (CES-D).
Prenatal 4-Month postpartum 12-Month postpartum 24-Month postpartum
N
M
SD
Percentage of mothers with 16 and higher score
248 219 219 197
16.44 12.07 11.05 10.12
9.17 9.48 9.00 9.52
47.2% (117) 25.7% (56) 22.9% (49) 22.8% (42)
in this paper, depressive symptom sums were dichotomized into two levels, ‘0 = not clinical level of depressive symptoms’ and ‘1 = clinical level depressive symptoms’ at each time point, using standard thresholds from the literature. The Neonatal Perception Inventory (NPI; Broussard & Hartner, 1970). The 12-item questionnaire measures mothers’ perception of child difficulty in their own child and the average child at prenatal and 1-day postnatal periods. The items ask mothers to rate newborns’ crying, feeding, vomiting, sleeping, and bowel movements in 5-point Likert scales, from ‘1 = none’ to ‘5 = a great deal,’ with a possible range of sum scores from 5 to 25 for both her own child and the average child, with a higher sum score indicating perception of greater infant difficulty. The design of the instrument allows one to subtract the mother’s rating of her own infant’s difficulty from the average infant’s difficulty, with a score of zero indicating she sees her child like the average child and with a more positive score indicating she sees her child as less difficult than the average child. The instrument was initially created to examine the relation between mothers’ perceptions of their children at birth and children’s emotional and developmental outcomes at later periods of life from childhood (Broussard & Hartner, 1970; Broussard and Hartner, 1971) to adulthood (Broussard & Cassidy, 2010). There is some precedent in the literature for adapting this instrument for prenatal period (Vaughn, Deinard, & Egeland, 1980; Minnesota Mother–Infant Interaction Project) as well as greater precedent for adapting mother-report measures of infant temperament for use during pregnancy (e.g., Contreras, Rhodes, & Mangelsdorf, 1995; Diener, Goldstein, & Mangelsdorf, 1995; Mebert, 1989; Zeanah et al., 1985). Such adaptations are based on a theoretical and empirical literature documenting that mothers begin to develop representations of their infants during pregnancy that form the foundation for representations after birth (Lashley et al., 2009; Zeanah et al., 1985). The Parental Stress Index Short Form (PSI; Abidin, 1995). Two subscales from this instrument were used to assess parental perception of child difficulty and difficult parent–child interactions The PSI conceptualizes parental stress as a system of factors that includes not only a mother’s experience of distress but also her perceptions of the child and her relationships with her child. The present study uses the ten items that measure the mother’s perception of the child as difficult (e.g., “my child seems to cry and fuss more than most children,” “my child gets upset easily over the smallest thing”) and the eleven items that measure the mother’s perception of their interaction as problematic (e.g., “my child smiles at me less than I expected,” “sometimes my child does things that bother me just to be mean”). The items measuring parental distress were not used. The response value scales are 5-point Likert scale, from ‘1 = Strongly Agree’ to ‘5 = Strongly Disagree.’ 2.2.2. Analysis plan After the preliminary and descriptive analyses of the data, the effect of prenatal depression on the longitudinal data of maternal perception was carried out using a linear mixed model (LMM) approach. LMM was selected because it accommodated missing data better than repeated measure analysis of variance, and it was considered by many experts as an appropriate way to examine multiple time-point longitudinal data (Krueger & Tian, 2004; Schafer & Graham, 2002). The analyses for the two NPI scales and the two PSI scales were carried out separately to accommodate the different rating norms for each questionnaire. Exploratory analyses were first conducted to identify demographic covariates that were correlated with depression and/or maternal perceptions. The potential covariates initially included mother’s age, education, child’s gender, and child’s treatment in a hospital neonatal intensive care unit after birth, but the final mixed models only included demographic covariates that were statistically significantly related to prenatal depression or maternal perceptions in bivariate analyses. The final mixed models also controlled for previous depressive symptom status (0 = below clinical threshold; 1 = above clinical threshold) at time points in addition to pregnancy. Analyses were carried out with the software package Stata version 12 (StataCorp, 2011). 3. Results 3.1. Descriptive analysis Prenatal depressive symptoms were high among the young mothers in the sample (see Table 2), with approximately 47% (N = 109) reporting symptoms levels above the clinical threshold for depressive symptoms with the total score of 16 and higher. The postpartum rate for clinical-level depressive symptoms continued to be relatively high but gradually decreased over time, with 25.7% (N = 56) at 4 months, 22.9% (N = 49) at 12 months, and 22.8% (N = 42) at 24 months (Edwards et al., 2012). The mothers’ prenatal level of depressive symptoms was highly correlated with their depressive symptoms postpartum at 4-months (r = 0.44, p < 0.001), 12-months (r = 0.48, p < 0.001), and 24-months (r = 0.53, p < 0.001).
H.Y. Lee, S.L. Hans / Infant Behavior & Development 40 (2015) 183–192
187
Table 3 Means, standard deviations, ANOVAs for neonatal perception inventory (NPI) and parenting stress inventory (PSI) scores by mother prenatal depression. Depressed
Not depressed
Variables
M
SD
M
SD
NPI Prenatal (N = 247) Your baby Average baby Differencea
F
p
13.90 16.41 2.51
3.51 3.09 2.80
13.06 15.25 2.19
2.83 2.73 2.18
4.28 9.75 1.02
0.04* 0.00*** 0.31
Postpartum (N = 235) Your baby Average baby Difference
19.20 15.61 3.53
3.68 3.72 3.15
17.45 14.23 3.22
3.01 2.96 2.79
10.00 16.09 0.62
0.00** 0.00*** 0.43
PSI 4 Month (N = 221) Difficult interaction Difficult child
17.20 23.43
4.93 6.39
15.43 21.22
3.49 6.39
9.63 6.61
0.00*** 0.01**
12 Month (N = 218) Difficult interaction Difficult child
17.49 25.91
4.99 6.96
15.66 21.88
4.10 6.15
8.93 20.47
0.00*** 0.00***
24 month (N = 197) Difficult interaction Difficult child
19.29 29.40
5.58 7.99
17.01 25.98
4.58 7.70
9.94 9.31
0.00*** 0.00***
a
Difference is calculated by subtracting your baby score from average baby score. p < 0.05. ** p < 0.01. *** p < 0.001. *
3.2. Study question 1: Comparison of maternal perceptions of prenatally depressed and non-depressed women Mothers’ perceptions of their children were examined prenatally and 1-day after childbirth, using the NPI, and then at 4-, 12-, and 24-months using the PSI. Table 3 summarizes the NPI and PSI scores by mother’s prenatal depressive symptom status. Prenatally depressed mothers anticipated more difficulty with their own and average infants at the prenatal and all of the postpartum periods than did mothers not depressed during pregnancy (p values ranging between
Contents lists available at ScienceDirect
Infant Behavior and Development
Full length article
Prenatal depression and young low-income mothers’ perception of their children from pregnancy through early childhood Helen Y. Lee a,∗ , Sydney L. Hans b a b
Boston University, Boston MA, United States University of Chicago, Chicago, IL, United States
a r t i c l e
i n f o
Article history: Received 27 August 2014 Received in revised form 25 February 2015 Accepted 2 June 2015 Available online 10 July 2015 Keywords: Early childhood Longitudinal data Mental representation Prenatal depression Perception Pregnancy
a b s t r a c t This study investigates the role of prenatal depression on young mothers’ perception of difficulty in child behavior and parent–child interaction from pregnancy through the first two years of child development. 248 low-income, African American women aged between 13 and 21 years reported on their perceptions of child behavior, parent–child interaction, and on depressive symptoms at the third trimester of pregnancy and at 4-, 12- and 24months postpartum. During pregnancy, a high percentage (47%) of the young mothers scored above the clinical level of prenatal depressive symptoms. These mothers anticipated and perceived significantly more child difficulty and parent–child interaction difficulty than did non-depressed mothers during pregnancy and over the first two years of the child’s development. Moreover, prenatal depression uniquely predicted negative maternal perception throughout the early years of child development even after adjusting for postpartum depression history. The enduring association between prenatal depression and a mother’s perception of her child is discussed with respect to the importance of pregnancy in mothers’ developing mental schema about their children and the emerging parent–child relationship. © 2015 Elsevier Inc. All rights reserved.
1. Introduction Depression during pregnancy is prevalent among women in the United States, affecting up to 38% of all pregnant women (Anderson, Sundstrom-Poromaa, Wulff, Astrom, & Bixo, 2003; Records & Rice, 2007) and 60% of at-risk mothers, including low-income teenage mothers (Earls & The Committee on Psychosocial Aspects of Child & Family Health, 2010; Goyal, Gay, & Lee, 2010). In recent years, the adverse effects of prenatal depression on maternal health and child development have received increasing attention in the literature, shedding light on its association with poor early mother–child interaction and child outcomes (for review see Field, Diego, & Hernandez-Reif, 2006) and its association with postpartum depression (Larsson, Sydsjo, & Josefsson, 2004; Milgrom et al., 2008). Much of the literature on prenatal depression to date has focused on the immediate and direct influence of maternal prenatal depression on the child via biochemical pathways in the womb environment (e.g., Lester, Conradt, & Marsit, 2013). However, yet another major way in which prenatal depression may impact the child after birth, which has not been
∗ Corresponding author. E-mail address: [email protected] (H.Y. Lee). http://dx.doi.org/10.1016/j.infbeh.2015.06.008 0163-6383/© 2015 Elsevier Inc. All rights reserved.
184
H.Y. Lee, S.L. Hans / Infant Behavior & Development 40 (2015) 183–192
well-explored, is through maternal cognition. The negative cognitions that depressed women establish about their unborn child during pregnancy may be patterns of thought that are the basis for future views of their children and their parent–child relationship, as well as their parenting behavior. A growing theoretical and empirical literature has highlighted the importance of parents’ representations of their children as a central feature of the parent–child relationship (e.g., Bugental & Happaney, 2002; Sigel, McGillicuddy-DeLisi, & Goodnow, 2014; Stern, 1989). Such representations are complex cognitive schemas of the child that are linked to emotions and to representations of self and others in the parent’s experience. Maternal representations of the child have been shown to be linked to maternal behavior, to infant attachment, and to infant development more generally (e.g., Broussard & Cassidy, 2010; Nover, Shore, Timberlake, & Greenspan, 2009; Zeanah & Barton, 1989). In particular, negative-valenced representations of the child, such as children being difficult, have been linked to developmental risk (Broussard & Cassidy, 2010). Mothers’ representations of the child begin before the birth of the child, and pregnancy is a crucial time when women anticipate and actively form ideas about their unborn children and about the impending parent–child relationship that forms the basis for enduring beliefs about the child (Ammaniti, Tambelli, & Odorisio, 2013; Lashley, Hans, & Henson, 2009; Nelson, 2003; Pajulo, Savonlahti, Sourander, Helenius, & Piha, 2001; Theran, Levendosky, Bogat, & Huth-Bocks, 2005; Zeanah, Keener, Stewart, & Anders, 1985). Depressive symptoms are closely linked to the ways mothers perceive and think about their children and their relationships with the children (Goodman & Gotlib, 1999; Joiner & Coyne, 1999). Depressed individuals show generally negative cognitive biases when processing social and affective information from interpersonal contexts (see review Bistricky, Ingram, & Atchley, 2011). They tend to selectively attend to and recollect negative or incongruent materials (Beck & Perkins, 2001; Gotlib, Krasnoperova, Yue, & Joormann, 2004), ruminate on negative information, and show more difficulty disengaging attention from negatively valenced materials than do non-depressed individuals (see review Gotlib & Joormann, 2010). As parents, women with depressive symptoms perceive more difficulty and problems with diverse aspects of their children and parent–child interactions across infancy to childhood (Ingersoll & Eist, 1998; Lovejoy, Graczyk, O’Hare, & Neuman, 2000; Weissman & Paykel, 1974). For example, mothers with depression perceive more difficulty in their newborns’ basic regulatory behavior, such as eating and bowel movements (Fulton, Mastergeoge, Steele, & Hansen, 2012), in temperamental characteristics of infants related to irritability and fussiness (McGrath, Records, & Rice, 2008), and in conduct problems during late childhood and adolescence (Conrad & Hammen, 1989; Galler, Harrison, Ramsey, Butler, & Forde, 2004; Teti, Gelfand, & Pompa, 1990; Whiffen & Gotlib, 1989). Despite the high prevalence of prenatal depression and the saliency of emerging parental cognition about the child and the parent–child relationship during pregnancy, the implication of depression during pregnancy on mothers’ immediate and longer-term perceptions of their children has been little examined. Some data suggest that prenatally depressed mothers may have a less strong emotional connection toward their fetus than other women (Lindgren, 2001; McFarland et al., 2011; Scharfe, 2007), but we know little about how prenatal depression affects women’s perceptions of their unborn children or predictions for what their children will be like after birth. Several studies suggest that prenatal maternal depression is associated longitudinally with mothers’ perceptions after birth of their infants as being difficult to care for. McGrath et al. (2008) found that prenatal depression was associated with reports of difficult temperament at two and six months after birth, although they did not control for maternal postnatal depression. Huot, Brennan, Stowe, Plotshky and Walker (2004) showed that prenatal maternal depression and not postnatal maternal depression related to mothers’ ratings of infant negative affect expression at six months of age. Similarly, Della Vedova (2014) found that women who are depressed during pregnancy rated their infants’ temperament as more difficult at three months after birth. A clinical sample of women with prenatal major depressive disorder were also relatively unlikely at three months to have balanced representations of their infants—characterized by positive, rich, and full descriptions of their infants (Wood, Hargreaves, & Marks, 2004). There are two main aims for the present study. First, the study aims to describe prenatally depressed mothers’ perceptions of child difficulty from the prenatal period through the first two years of child development, including the first hours after birth, and to compare their perceptions to those of non-depressed mothers. Little is currently known about the trajectory of changes in perception between depressed and non-depressed mothers from pregnancy to postpartum, especially in the earliest days of the neonatal period when mothers have not had an opportunity to revise their ideas about the child based on interaction with the child. Second, the study aims to examine the contribution of prenatal depression on mothers’ perception of their children over time by taking later depressive symptoms into account. Although the independent effect of prenatal depression after adjusting for postpartum depression has been previously found with respect to later child outcomes (Deave, Heron, Evans, & Emond, 2008; Pearson et al., 2012; Hay, Pawlby, Waters, Perra, & Sharp, 2010), it has not been examined with respect to maternal perceptions of the child. Moreover, considering that previous literature has suggested that the history of depression is a major predictor of mothers’ postpartum depression (for review Lancaster et al., 2010), the present study was designed to adjust for the prior postpartum depression status in addition to mothers’ concurrent depression status. 2. Methods 2.1. Participants 248 young women were recruited from two prenatal care clinics affiliated with a large Midwestern university hospital system as a part of a larger intervention study. The women were eligible to participate in the study if they were under 22
H.Y. Lee, S.L. Hans / Infant Behavior & Development 40 (2015) 183–192
185
Table 1 Demographic characteristics of participants. Characteristics
M
SD
Mother age in years (N = 248) Highest education level completed in years (N = 247)
17.6 10.7 n
1.7 1.6 %
30 22 17 142 18 2 1
12.9 9.5 7.3 61.2 7.8 0.9 0.4
218 29 1
87.5 12.1 0.4
128 115
52.7 47.3
174 71
71.0 29.0
Relationship with father of baby (N = 232) No relationship Some level of friendship Good friends Boyfriend Fiancé Married Other Participant’s other children (N = 248) 0 1 2 Gender of baby (N = 243) Boy Girl Neonatal Intensive Care Unit (NICU) experience (N = 245) No Yes
years of age, less than 34 weeks gestation, not planning to move from the area, and willing to participate in an intervention study offering health and childbirth education and support. All of the young mothers recruited from the clinics were AfricanAmerican. At enrollment, the mothers’ ages ranged between 13 and 21 years, with a mean age of 17.6 (SD = 1.7) years. Most mothers received Medicaid (94% or N = 233). Mothers had completed an average of 10.7 years of school (SD = 1.6), and 37.9% (N = 94) had completed high school. 13% (N = 30) of the women reported having no relationship with the father of their baby during pregnancy, 86% (N = 201) reported having an ongoing friendship or romantic relationship with the father of their baby, and 1% (N = 2) reported being married. 88% of the women (N = 203) were first-time mothers. The mean weight of the children was 6.80 pounds (SD = 1.32) at birth. 29% of the children (N = 71) were housed in the hospital Neonatal Intensive Care Unit (NICU) after birth, rather than allowed to room in with their mother because of minor to severe medical problems with respect to the infant or mother. Table 1 shows detailed demographic information. The mothers were interviewed at five time points: pregnancy (mean 84 days before giving birth), 1 day after the birth, and 4-, 12-, and 24-months after the birth. 90.7% (N = 235) of the sample was retained for interviews at 1 day, 89.1% (= 221) at 4 months, 88.3% (n = 219) at twelve months and 79.4% (n = 219) at 12 months and 79.4% ( = 197) at 24 months. Reasons for loss of data at one day were primarily mothers giving birth at non-study hospitals or maternal and infant medical complications that made interviewing difficult. At later waves, data loss was primarily due to inability to locate mothers or to schedule interviews within an appropriate time frame, although 3 infants died as newborns, 1 mother withdrew from study participation, and 3 mothers lost or surrendered custody of their infants. Some mothers who were lost at one wave returned to the sample at later longitudinal assessment points. Analyses were conducted to examine potential differential loss to follow up based on a set of 20 demographic, medical, and psychosocial characteristics measured at enrollment. Retention was related to only one of these baseline variables: mothers who had been working at enrollment were less likely to be participating in the study at 4, 12, and 24 months. Mothers who had high prenatal depressive symptoms were not significantly more likely to be lost to follow up at 4, 12 or 24 months (12.0%, 14.5%, and 23.1%, respectively) than were mothers with low prenatal depressive symptoms (11.5%, 9.2%, 17.6%). 2.2. Procedure Maternal depressive symptoms were assessed with the Center for Epidemiology Studies Depression Scale (CES-D) at each time point, except at 1 day after birth. Maternal perceptions of the infant prenatally and 1 day postpartum were measured by the Neonatal Perception Inventory (NPI). Mothernal perceptions of the older infants were measured when the child was 4-, 12-, and 24-months old by the Parenting Stress Index (PSI). Demographic and medical covariates were assessed prenatally or neonatally. 2.2.1. Measures Center for Epidemiological Studies Depression Scale (CES-D; Radloff, 1977). This self-report 20-item questionnaire surveys mood, somatic complaints, interactions with others, and motor functioning. The response values are 4-point Likert scales, with a range of 0-3, with ‘0 = rarely or none of the time (less than one day)’ to ‘3 = most or all of the time (5-7days).’ The final sum score can be anywhere between 0 and 60, with a higher score indicating greater impairment, and classifies a total score of 16 or higher as a depressive case according to the instrument’s authors. For the purpose of all statistical analysis
186
H.Y. Lee, S.L. Hans / Infant Behavior & Development 40 (2015) 183–192
Table 2 Means and standard deviations for maternal depressive symptom scores (CES-D).
Prenatal 4-Month postpartum 12-Month postpartum 24-Month postpartum
N
M
SD
Percentage of mothers with 16 and higher score
248 219 219 197
16.44 12.07 11.05 10.12
9.17 9.48 9.00 9.52
47.2% (117) 25.7% (56) 22.9% (49) 22.8% (42)
in this paper, depressive symptom sums were dichotomized into two levels, ‘0 = not clinical level of depressive symptoms’ and ‘1 = clinical level depressive symptoms’ at each time point, using standard thresholds from the literature. The Neonatal Perception Inventory (NPI; Broussard & Hartner, 1970). The 12-item questionnaire measures mothers’ perception of child difficulty in their own child and the average child at prenatal and 1-day postnatal periods. The items ask mothers to rate newborns’ crying, feeding, vomiting, sleeping, and bowel movements in 5-point Likert scales, from ‘1 = none’ to ‘5 = a great deal,’ with a possible range of sum scores from 5 to 25 for both her own child and the average child, with a higher sum score indicating perception of greater infant difficulty. The design of the instrument allows one to subtract the mother’s rating of her own infant’s difficulty from the average infant’s difficulty, with a score of zero indicating she sees her child like the average child and with a more positive score indicating she sees her child as less difficult than the average child. The instrument was initially created to examine the relation between mothers’ perceptions of their children at birth and children’s emotional and developmental outcomes at later periods of life from childhood (Broussard & Hartner, 1970; Broussard and Hartner, 1971) to adulthood (Broussard & Cassidy, 2010). There is some precedent in the literature for adapting this instrument for prenatal period (Vaughn, Deinard, & Egeland, 1980; Minnesota Mother–Infant Interaction Project) as well as greater precedent for adapting mother-report measures of infant temperament for use during pregnancy (e.g., Contreras, Rhodes, & Mangelsdorf, 1995; Diener, Goldstein, & Mangelsdorf, 1995; Mebert, 1989; Zeanah et al., 1985). Such adaptations are based on a theoretical and empirical literature documenting that mothers begin to develop representations of their infants during pregnancy that form the foundation for representations after birth (Lashley et al., 2009; Zeanah et al., 1985). The Parental Stress Index Short Form (PSI; Abidin, 1995). Two subscales from this instrument were used to assess parental perception of child difficulty and difficult parent–child interactions The PSI conceptualizes parental stress as a system of factors that includes not only a mother’s experience of distress but also her perceptions of the child and her relationships with her child. The present study uses the ten items that measure the mother’s perception of the child as difficult (e.g., “my child seems to cry and fuss more than most children,” “my child gets upset easily over the smallest thing”) and the eleven items that measure the mother’s perception of their interaction as problematic (e.g., “my child smiles at me less than I expected,” “sometimes my child does things that bother me just to be mean”). The items measuring parental distress were not used. The response value scales are 5-point Likert scale, from ‘1 = Strongly Agree’ to ‘5 = Strongly Disagree.’ 2.2.2. Analysis plan After the preliminary and descriptive analyses of the data, the effect of prenatal depression on the longitudinal data of maternal perception was carried out using a linear mixed model (LMM) approach. LMM was selected because it accommodated missing data better than repeated measure analysis of variance, and it was considered by many experts as an appropriate way to examine multiple time-point longitudinal data (Krueger & Tian, 2004; Schafer & Graham, 2002). The analyses for the two NPI scales and the two PSI scales were carried out separately to accommodate the different rating norms for each questionnaire. Exploratory analyses were first conducted to identify demographic covariates that were correlated with depression and/or maternal perceptions. The potential covariates initially included mother’s age, education, child’s gender, and child’s treatment in a hospital neonatal intensive care unit after birth, but the final mixed models only included demographic covariates that were statistically significantly related to prenatal depression or maternal perceptions in bivariate analyses. The final mixed models also controlled for previous depressive symptom status (0 = below clinical threshold; 1 = above clinical threshold) at time points in addition to pregnancy. Analyses were carried out with the software package Stata version 12 (StataCorp, 2011). 3. Results 3.1. Descriptive analysis Prenatal depressive symptoms were high among the young mothers in the sample (see Table 2), with approximately 47% (N = 109) reporting symptoms levels above the clinical threshold for depressive symptoms with the total score of 16 and higher. The postpartum rate for clinical-level depressive symptoms continued to be relatively high but gradually decreased over time, with 25.7% (N = 56) at 4 months, 22.9% (N = 49) at 12 months, and 22.8% (N = 42) at 24 months (Edwards et al., 2012). The mothers’ prenatal level of depressive symptoms was highly correlated with their depressive symptoms postpartum at 4-months (r = 0.44, p < 0.001), 12-months (r = 0.48, p < 0.001), and 24-months (r = 0.53, p < 0.001).
H.Y. Lee, S.L. Hans / Infant Behavior & Development 40 (2015) 183–192
187
Table 3 Means, standard deviations, ANOVAs for neonatal perception inventory (NPI) and parenting stress inventory (PSI) scores by mother prenatal depression. Depressed
Not depressed
Variables
M
SD
M
SD
NPI Prenatal (N = 247) Your baby Average baby Differencea
F
p
13.90 16.41 2.51
3.51 3.09 2.80
13.06 15.25 2.19
2.83 2.73 2.18
4.28 9.75 1.02
0.04* 0.00*** 0.31
Postpartum (N = 235) Your baby Average baby Difference
19.20 15.61 3.53
3.68 3.72 3.15
17.45 14.23 3.22
3.01 2.96 2.79
10.00 16.09 0.62
0.00** 0.00*** 0.43
PSI 4 Month (N = 221) Difficult interaction Difficult child
17.20 23.43
4.93 6.39
15.43 21.22
3.49 6.39
9.63 6.61
0.00*** 0.01**
12 Month (N = 218) Difficult interaction Difficult child
17.49 25.91
4.99 6.96
15.66 21.88
4.10 6.15
8.93 20.47
0.00*** 0.00***
24 month (N = 197) Difficult interaction Difficult child
19.29 29.40
5.58 7.99
17.01 25.98
4.58 7.70
9.94 9.31
0.00*** 0.00***
a
Difference is calculated by subtracting your baby score from average baby score. p < 0.05. ** p < 0.01. *** p < 0.001. *
3.2. Study question 1: Comparison of maternal perceptions of prenatally depressed and non-depressed women Mothers’ perceptions of their children were examined prenatally and 1-day after childbirth, using the NPI, and then at 4-, 12-, and 24-months using the PSI. Table 3 summarizes the NPI and PSI scores by mother’s prenatal depressive symptom status. Prenatally depressed mothers anticipated more difficulty with their own and average infants at the prenatal and all of the postpartum periods than did mothers not depressed during pregnancy (p values ranging between
