Maternal Depression Trajectories Relate to Youths’ Psychosocial and Cognitive Functioning at Adolescence and Young Adulthood

Abstract

This study evaluated how patterns of mothers’ depressive symptoms across their child’s childhood relate to children’s psychosocial adjustment at adolescence and young adulthood and to cognitive functioning at adolescence. Depressive symptoms were measured in 1,273 mothers when their children were 1, 5, 10, and 14.6 years of age. Children (53.5% male; n = 1,024) completed the Youth Self-Report at adolescence (M = 14.6y), and the Adult Self-Report in young adulthood (M = 20.5y; n = 817) to assess internalizing and externalizing symptoms. Adolescents also completed standardized cognitive tests to assess verbal and mathematical skills. Growth mixture modeling analyses identified four patterns of maternal depressive symptom trajectories: infrequent (55%), increasing at adolescence (20%), decreasing at adolescence (14%), and chronic severe (11%). Results indicated that exposure to maternal depression of any duration, severity or time period during childhood portended higher levels of externalizing and attention problems at both adolescence and adulthood and higher levels of internalizing problems at adulthood. Adolescents whose mothers had chronic severe depressive symptoms had lower language, vocabulary, reading comprehension and mathematical test scores than youth whose mothers had stable infrequent depressive symptoms. Findings illustrate the significance and long-term ramifications of mothers’ depressed mood for their children’s mental and psychosocial health into adulthood. Findings also demonstrate that the lower cognitive abilities among children of severely depressed mothers persist beyond childhood and pertain to a broad range of cognitive abilities.

Introduction

Maternal depression is a global public health issue that negatively affects the health and well-being of mothers and their children (Surkan et al., 2011; Wachs et al., 2009). Globally, rates of maternal depression vary by country income level, with prevalence estimates ranging from 7 to 15% in high-income countries and between 20 to 25% in low- and middle-income countries (Gelaye et al., 2016; Kessler & Bromet, 2015). The prevalence of maternal depression in Chile is quite high, with one-third to 50% of Chilean women experiencing depressive symptoms (Jadresic et al., 2007; Quelopana et al., 2011; Wolf et al., 2002). Factors contributing to depression among Chilean mothers involve low levels of education, having many children at young ages, poverty, high household occupancy, and living in multi-generational households (Jadresic et al., 2006; Quelopana et al., 2011). Patriarchal gender roles, with mothers singly responsible for the care and welfare of home, family and children, also likely contribute to mothers’ depression (Shidhaye & Giri, 2014).

For many mothers with depression, symptoms can persist. McCue Horwitz and colleagues determined that 46% of mothers with initial elevated depressive symptoms continued to have elevated symptoms 1 year later (McCue Horwitz et al., 2007). Such high persistence of depressive symptoms necessitates further exploration of their effects, not only for mothers, but also for their children (Goodman, 2007; Goodman et al., 2011). For example, studies have shown how exposure to across-time patterns of maternal depression affects children, examining effects related to the severity, chronicity and timing of mothers’ depressive symptoms ( Ahun et al., 2017; Ashman et al., 2008; Buckingham-Howes et al., 2017; Cents et al., 2013; Matijasevich et al., 2015). This research shows that persistent maternal depression, regardless of severity, relates to children’s emotional and behavioral problems (Foster et al., 2008; Kingston et al., 2018; van der Waerden et al., 2015), while other studies find that an increasing trend in mothers’ depressive symptoms is especially problematic for children’s adjustment (Flouri & Ioakeimidi, 2018; Guyon-Harris et al., 2016; Park et al., 2018). Several studies also find that children of mothers with chronic depressive symptoms fare most poorly (Ashman et al., 2008; Campbell et al., 2009; Wickham et al., 2015).

Most studies that have examined the effects of mothers’ across-time patterns of depression on children have focused on outcomes apparent in early childhood. Only a few studies have examined the effects of maternal depression on children’s adjustment out to adolescence (e.g., Campbell et al., 2009; Halligan,et al., 2007; Pearson et al., 2013; Wickham et al., 2015). No study of which we are aware has examined the effects of mothers’ depression at children’s young adulthood. Additionally, little research has examined mothers’ depressive patterns across their child’s entire childhood, with studies often contrasting mothers’ prenatal depression to immediate postpartum levels. Moreover, most studies have relied on mothers’ reports of their own depression as well as their child’s adjustment, creating shared-reporter bias.

Maternal depression has also been linked with children’s lower cognitive abilities. In a large study involving over 7,600 mother-child dyads, Petterson and Albers (2001) found that mothers’ level of depression was associated with children’s lower social, motor, and language development at 2 to 4 years of age. In examining timing effects, Ahun et al. (2017) found that children exposed to chronic maternal depression had lower vocabulary scores at age 10 than children never exposed, exposed early or exposed late. Mothers’ acute depressive symptoms can also have harmful effects for children’s early cognitive development (Campbell et al., 2007; Conners-Burrow et al., 2014). Little is known, however, about how the timing or chronicity of maternal depression relates to children’s cognitive abilities at adolescence, such as verbal or mathematical skills. Such findings would reveal enduring and long-term effects of mothers’ depression for their children’s cognitive abilities.

Several theories have been proposed for linking maternal depression to children’s emotional and behavioral problems. There is possible genetic transmission between mother and child of affective and psychobiological vulnerabilities for depression (Goodman, 2007; Sullivan et al., 2000), as well as shared environmental influences (e.g., poverty, poor nutrition) (Sullivan et al., 2000). Many explanations also suggest that children of depressed mothers are more likely to be exposed to hostile or withdrawn parenting and a range of negative interpersonal behaviors in their mothers, which could serve as models for maladaptive social behaviors (Field, 2010; Goodman, 2007; Hammen & Brennan, 2001; Lovejoy et al., 2000). Depressed mothers may also be less able to help and support their child if he or she is experiencing mental health issues themselves (Perloe et al., 2014). There is also some evidence of a concurrent escalation of mothers’ and adolescents’ depressive symptoms over time (Perloe et al., 2014), suggesting a transactional or bidirectional relation between mother and child.

In the current study, we examined how patterns of mothers’ depressive symptoms across their children’s childhood relate to youths’ psychosocial adjustment at adolescence and young adulthood. We also examined how the across-time changes in mothers’ depression relate to youths’ cognitive functioning at adolescence, specifically their verbal and mathematical abilities as assessed on standardized intelligence tests. Based on evidence that the emotional withdrawal and detachment stemming from depression lead mothers to be less warm and affectionate toward their children (i.e., a parenting explanation; Field, 2010; Lovejoy et al., 2000), we hypothesize that youth of mothers experiencing chronic and severe depression will evidence more problematic adjustment at both adolescence and young adulthood than youth whose mothers had infrequent depressive symptoms. In addition, drawing on studies that show that severely depressed mothers are less responsive and engaged with their children (Field, 2010; Lovejoy et al., 2000), we hypothesize that adolescents of mothers with chronic and severe depression will score lower on tests of verbal, reading, and mathematical abilities.

Methods

Sample and Study Design

Participants were mothers and their adolescent and young adult children studied as part of the Santiago Longitudinal Study (SLS), which began as a randomized-controlled iron-deficiency anemia preventive trial and neuromaturation study in Santiago, Chile (Lozoff et al., 2003). Healthy infants were recruited at 6 months of age from community clinics from four contiguous working-class communities, with 1,790 infants enrolled. Eligible infants were full-term singletons, weighing ≥ 3.0 kg at birth and had no major congenital anomalies, perinatal complications or chronic illness. Mothers who were illiterate or had another infant < 1 year of age were excluded. Recruitment occurred from 1991 to 1996. Children and mothers were studied at four time points: when children were 1 year, 5.5 years, 10 years, and between 12–17 years of age (M age = 14.6 years). (The adolescent assessment was part of a project that began in 2008, when children’s ages ranged from 12 to 17 years.) Only youth were followed up in young adulthood (M age = 20.5y; range: 20–23y). Mothers who had depressive symptom scores from at least two of the four study time points were included in analyses (N = 1,273). Neither mothers nor children received any special psychosocial services as part of the study.

Mothers included in the current analyses had slightly but significantly lower depression scores when their child was 1 year compared to those who were not studied. Adolescents included in the current analyses were more likely to be female, from families of higher socioeconomic status (SES) and had mothers with slightly higher IQs than those not analyzed here (measures described below in Covariates). Young adults included in the current analyses were more likely to be female, and had lower family stress scores at infancy. Analyses adjusted for these factors.

Approval for this study was obtained from the authors’ university Institutional Review Boards in the U.S. and Chile. Informed written consent was obtained from mothers for their and their child’s participation at the four study time points; assent was obtained from children at age 10 and in adolescence. Youth provided informed written consent for their participation at young adulthood. All study procedures were in accord with the code of ethics of the World Medical Association (World Medical Association, 2013).

Measures

Maternal Depressive Symptoms.

Mothers completed the 20-item Center for Epidemiologic Studies-Depression Scale (CES-D) (Radloff, 1977) when children were 1, 5.5, and 10 years, and between 12–17 years of age. This questionnaire asks about the frequency of depressed mood within the past week (“I could not get going”), with response options ranging from “rarely or none of the time” (0) to “most or all the time” (3). Items were summed so that scores ranged from 0 to 60. In clinical practice, scores 16–23 indicate mild to moderate depression, and scores 24–60 indicate severe depression (Radloff, 1977). In this study, we use the CES-D scores along a continuum to indicate severity, without diagnostic categorization. The Cronbach alphas ranged between .83 – .91 at the four study time points.

Adolescent Behavioral and Emotional Adjustment.

At adolescence, youth completed the Youth Self-Report Child Behavior Checklist (YSR) (Achenbach, 1991), which includes scales indexing internalizing symptoms (anxiety/depression, withdrawn/depression, somatic complaints), externalizing problems (rule-breaking, aggression), attention and social problems, and total problems (sum of problems from all seven scales) within the past 6 months. Response options are: 0 = “not true,” 1 = “somewhat or sometimes true,” and 2 = “very true or often true.” Scores were summed across items, with all scales having good internal reliability (Cronbach coefficient alphas: .63 – .81). Raw scores were converted to T scores (range: 50–100 and standardized for age and sex). T scores correspond to percentile scores, with T scores below 65 corresponding to percentile scores below the 95^th percentile and considered to be in the normal range. T scores of 65 to 69 correspond to percentile scores between the 95^th and 98^th percentiles and are considered to be in the borderline clinical range. T scores ≥ 70 correspond to percentile scores above the 98^th percentile and are considered to be in the clinical range (Achenbach, 1991). We analyze T scores along a continuum to indicate behavior problem severity; T scores ≥ 65 are not intended to indicate a diagnostic categorization.

Young Adult Behavioral and Emotional Adjustment.

At young adulthood, youth completed the Adult Self-Report (ASR) (Achenbach & Rescorla, 2003), which includes scales indexing internalizing and externalizing symptoms (analogous but not identical to the YSR scales), as well as attention problems within the past 6 months. Response options range from 0 (“not true”) to 2 (“very true or often true”). Raw scores were converted to T scores (range: 50 – 100, standardized by sex and age), which were analyzed along a continuum to indicate behavior problem severity and not a diagnostic categorization. Cronbach alphas of the various scales have been reported to range from .78 to .88 (Achenbach & Rescorla, 2003).

Adolescent Cognitive Functioning and Achievement.

Adolescents completed two subtests of the Wechsler Intelligence Scale for Children, 3rd ed. (WISC-III) (Wechsler, 1991): the language similarities subtest (a measure of verbal abilities) and the matrix reasoning subtest (a measure of quantitative reasoning). Standardized scores of both subtests were used (range 1 – 19). Adolescents also completed the Spanish Reading Test (El Examen de Lectura en Español), assessing vocabulary (range 0–45) and comprehension (range 0–30). In addition, adolescents completed the mathematics subtest of the Wide Range Achievement Test (WRAT) (Wilkinson & Robertson, 2006). Standardized scores were analyzed (mean 100, SD 15). Youth completed all tests between 15.3 and 17.4 years of age (M age 16.2 years; Table 1).

Table 1.

Description of Sample and Study Scores (N = 1,273)

Participant characteristics	Mean or %	SD	Range
Mothers’ age at child’s infancy (years)	26.4	6.1	16 – 44
Mothers’ age at child age 5 (years)	31.9	5.4	21 – 47
Mothers’ age at child age 10 (years)	36.6	4.4	26 – 52
Mothers’ age at child’s adolescence (years)	41.1	4.7	29 – 57
Mothers’ IQ	83.8	9.5	60 – 110
Mothers’ years of education	9.4	2.7	1 – 17
Fathers’ years of education	9.8	2.8	0 – 19
Family SESa	27.5	6.3	9 – 47
Number of children < 15y in household	2.1	1.2	0 – 12
Number of family stressors	4.7	2.7	0 – 14
Child age 1 (years)	1.1	0.3	0.9 – 1.2
Child age 5 (years)	5.5	0.5	5.4 – 6.0
Child age 10 (years)	10.0	0.1	10 – 11
Adolescent assessment (N=1,024)
Youth age (years)	14.6	1.4	12 – 17
Youth sex at adolescence (% male)	53.5%
Cognitive assessment at adolescence (N=955)
Youth age (years)	16.2	0.2	15 – 17
Youth sex (% male)	50.3%
WISC - language similarities	8.4	2.1	2 – 17
WISC - matrix reasoning	7.5	2.4	1 – 15
Reading – vocabulary	21.6	7.3	3 – 40
Reading – comprehension	11.7	4.9	1 – 29
WRAT – mathematics	82.4	10.1	45 – 119
Young adult assessment (N=817)
Youth age (years)	20.5	0.6	20 – 23
Youth sex (% male)	45%

Note.

^aFamily socioeconomic status was scored such that higher scores indicate greater disadvantage.

Study covariates.

Participant characteristics analyzed as covariates were youth sex, youth age, mothers’ years of education, maternal IQ (assessed by the Wechsler Adult Intelligence Scale; Wechsler, 1955), family SES, and family stress. Family SES was assessed using the 13-item Graffar poverty index, which assesses family living and housing conditions and material possessions (Graffar, 1956). Family stress was assessed using the 30-item Holmes and Rahe (1967) social readjustment scale, which indexes the number of stressors affecting the family (moves, marital separations, etc.). The total number of stressors was used in analyses.

Statistical Analyses

We used growth mixture modeling (GMM; Mplus 8.2) with maximum likelihood estimation to model maternal depression trajectories at the four study time points (Nagin, 2005). Data from mothers who had CES-D scores from at least two of the four study time points were analyzed (N=1,273) (Schafer & Graham, 2002). All cases were retained using the full information maximum likelihood (FIML), which is a recommended method that fits the model being tested directly onto the non-missing data for each participant (Enders & Bandalos, 2001). To assess the optimal number of trajectories, GMM specified 1-, 2-, 3-, 4- and 5-group solutions. The Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC) scores were compared for each model, with both indices improving successively with each added class (Supplemental Table 1). We adopted the 4-class solution based on multiple fit statistics (Nagin, 2005). The average posterior probability of group membership per latent class was slightly higher for the 4-class model compared to the 5-class model (.79 vs. .75, respectively); both the 3-class and 5-class models had probabilities < .68 for some groups. (The average posterior probability of trajectory membership should be at least > .70; Nagin, 2005). In addition, the 4-class model had more disparate maternal CES-D scores at children’s adolescence, with two groups having CES-D scores in the non-clinical range and two groups having CES-D scores in the clinical range. The 5-class model had only one group with an average CES-D score < 16 at adolescence, with the other 4 groups having high CES-D scores at adolescence. Additionally, the 5-class solution had one group with a small sample size (n = 69).

We then compared youths’ YSR and ASR T-scores by maternal depression trajectory group using multivariate analysis of covariance (MANCOVA), controlling for covariates (SPSS v. 26). MANCOVA F values determined maternal depression group differences across all psychosocial scores, and follow-up analysis of covariance (ANCOVA) was computed on individual T scores, using Bonferroni correction for multiple group comparisons. Adolescents’ cognitive test scores were compared across maternal depression trajectory group using ANCOVA, with Bonferroni correction for multiple group comparisons. Participant characteristics analyzed as covariates were youth sex, youth age, mothers’ years of education, maternal IQ, family SES, and family stress. All analyses adjusted for these factors.

Results

Sample Description

Mothers were an average age of 26.4 years (SD = 6.1) at their child’s infancy, and 31.9, 36.6, and 41.1 years at the other study time points. Mothers’ had an average of 9.4 years of education (only 9 years of education was compulsory in Chile at the time of the study) (Table 1). At adolescence, 1,024 youth had adjustment scores (53.5% male), 955 had cognitive scores (50.3% male), and 817 had adjustment scores at young adulthood (45% male).

Growth Trajectories

A 4-group solution provided the best-fitting model (Supplemental Table 1). Fit indices of the growth mixture model were: CFI = .926, RMSEA = .018, SRMR = .017 (intercept and slope for each trajectory group are shown in Supplementary Table 2). Mothers’ mean CES-D scores per each trajectory group are illustrated in Figure 1 and shown in Table 2. Class 1 (“infrequent”; n = 698, 55%) included mothers who consistently had infrequent depressive symptoms. Class 2 mothers (“increasing at adolescence”; n = 251, 20%) had borderline-high depressive symptoms at child ages 1, 5.5 and 10, with scores increasing substantially from child age 10 years to adolescence. Class 3 mothers (“decreasing at adolescence”; n = 180, 14%) had steadily increasing depressive symptoms from child age 1 to 10, with symptoms decreasing dramatically from child age 10 years to adolescence. Class 4 mothers had consistently very high depressive symptoms throughout their child’s development (“chronic severe”; n = 144, 11%).

Figure 1.

Growth trajectories showing four maternal depressive symptom classes from child age 1 year to adolescence (n = 1,273). Class 1 = Infrequent depressive symptoms. Class 2 = Increasing depressive symptoms (at child’s adolescence). Class 3 = Decreasing depressive symptoms (at child’s adolescence). Class 4 = Chronic severe depressive symptoms. CES-D: Center for Epidemiological Studies-Depression Inventory.

Table 2.

Mothers’ Depression Scores by Trajectory Class

	Class 1 Infrequent (n =698)		Class 2 Increasinga (n=251)		Class 3 Decreasingb (n=180)		Class 4 Chronic severe (n=144)
	M	(SD)	M	(SD)	M	(SD)	M	(SD)
Child age 1y	11.1	(8.9)	19.4	(118)	18.0	(117)	29.6	(119)
Child age 5.5y	14.0	(10.9)	23.7	(12.5)	24.9	(12.6)	36.5	(10.1)
Child age 10y	9.2	(6.2)	17.9	(6.9)	32.4	(6.5)	39.9	(6.5)
Child age 14y	10.2	(7.0)	34.8	(7.2)	15.8	(7.3)	40.4	(7.8)

Note. Scores were derived from the Center for Epidemiological Studies-Depression Scale (CES-D).

^aThe ‘increasing’ trajectory reflects increasing depressive symptoms at child’s adolescence.

^bThe ‘decreasing’ trajectory reflects decreasing depressive symptoms at child’s adolescence.

Psychosocial and Behavioral Adjustment at Adolescence by Maternal Depression Trajectory

The MANCOVA test indicated that youths’ YSR T scores at adolescence differed significantly by maternal depression trajectory (F [30, 2868] = 1.84, P = .004; Table 3). Follow-up ANCOVA results showed that youth whose mothers had chronic severe depressive symptoms (class 4) had the most problematic internalizing scores than any other group (i.e., for YSR scales of anxious-depressed, withdrawn-depressed, somatic complaints, and total internalizing symptoms). Youth in the increasing and decreasing maternal depression trajectory groups (classes 2 and 3) and youth in the infrequent and decreasing groups (classes 1 and 3) were equivalent for all internalizing symptoms. However, compared to youth whose mothers experienced infrequent depression (class 1), youth whose mothers had increasing pattens of depression (class 2) reported more somatic complaints and total internalizing symptoms.

Table 3.

Estimated Mean T scores of Adolescents’ Psychosocial and Behavioral Adjustment by Mothers’ Depression Trajectory

Adolescents’ T-scores	Class 1 Infrequent (n=544) M (SE)	Class 2 Increasinga (n=209) M (SE)	Class 3 Decreasingb (n=155) M (SE)	Class 4 Chronic severe (n=116) M (SE)	F (3, 986)	P	Class contrasts	Equivalent classes
Anxious-depressed	56.65a (0.29)	57.72 (0.47)	57.62 (0.54)	58.64a (0.63)	3.41	.017	1 < 4	1 = 2 = 3 2, 3 = 4
Withdrawn-depressed	57.16a (0.32)	57.71b (0.51)	57.98c (0.59)	60.05a,b,c (0.69)	4.77	.003	1, 2, 3 < 4	1 = 2 = 3
Somatic complaints	54.74a,b (0.26)	55.90a (0.41)	54.95c (0.47)	56.74b,c (0.55)	4.67	.003	1< 2, 4; 3 < 4	1 = 3; 2 = 3 2 = 4
Internalizing problems	54.70a,b (0.38)	56.16a,c (0.60)	55.72d (0.69)	58.23b,c,d (0.81)	5.55	.001	1 < 2 < 4; 3 < 4	1 = 3; 2 = 3
Rule-breaking behaviors	55.93a,b,c (0.26)	57.07a (0.42)	57.03b (0.49)	58.00c (0.57)	4.61	.003	1 < 2, 3, 4	2 = 3 = 4
Aggressive behaviors	56.19a,b,c (0.31)	57.62a (0.50)	57.51b (0.58)	59.18c (0.68)	6.19	< .001	1 < 2, 3, 4	2 = 3 = 4
Externalizing behaviors	53.93a,b,c (0.40)	56.05a (0.64)	56.15b (0.74)	57.47c (0.86)	6.56	< .001	1 < 2, 3, 4	2 = 3 = 4
Attention problems	55.17a,b,c (0.29)	56.80a (0.46)	56.95b (0.53)	57.10c (0.62)	5.68	.001	1 < 2, 3, 4	2 = 3 = 4
Social problems	56.14 (0.28)	57.46 (0.44)	56.77 (0.51)	57.14 (0.60)	2.38	.068
Total problems	53.68a,b,c (0.37)	55.80a (0.59)	55.26b (0.68)	57.23c (0.80)	6.99	< .001	1 < 2, 3, 4	2 = 3 = 4

Note. Estimated mean values with the same letter superscript in the same row are significantly different (P < .05). Analyses adjusted for adolescent sex, age, mother education, maternal IQ, family SES, and family stress.

^aThe ‘increasing’ trajectory reflects increasing depressive symptoms at child’s adolescence.

^bThe ‘decreasing’ trajectory reflects decreasing depressive symptoms at child’s adolescence.

Results further showed that youth whose mothers had infrequent depressive symptoms (class 1) had significantly lower externalizing (rule-breaking, aggression), attention and total problems than all other groups (Table 3). Youth whose mothers had increasing, decreasing, or chronic severe depression trajectories (classes 2, 3, or 4) were equivalent in externalizing symptoms (rule-breaking, aggression), attention problems, and total problems. No group differences were apparent for youths’ social problems.

Psychosocial and Behavioral Adjustment at Young Adulthood by Maternal Depression Trajectory

The MANCOVA test on young adults’ ASR T scores (Table 4) indicated that young adults’ psychosocial problems differed significantly by maternal depression trajectory (F [27, 2167] = 1.60, P = .026). Follow-up ANCOVA results showed that young adults’ whose mothers had infrequent depressive symptoms (class 1) had fewer psychosocial problems across all problems analyzed than young adults in the other groups. Young adults whose mothers experienced increasing, decreasing, or chronic severe depressive patterns throughout their child’s childhood (classes 2, 3, and 4) reported relatively equivalent levels of psychological and behavioral problems as young adults. In addition, young adults whose mothers experienced infrequent depression (class 1) experienced comparable levels of somatic complaints and rule-breaking behaviors as those whose mothers showed decreasing depressive trajectories (class 3).

Table 4.

Estimated Mean T scores of Young Adults’ Psychosocial and Behavioral Adjustment by Mothers’ Depression Trajectory

Young adults’ T-scores	Class 1 Infrequent (n=448) M (SE)	Class 2 Increasinga (n=171) M (SE)	Class 3 Decreasingb (n=115) M (SE)	Class 4 Chronic severe (n=83) M (SE)	F (3, 750)	P	Class contrasts	Equivalent classes
Anxious-depressed	57.65a,b,c (0.36)	59.25a (0.58)	59.04b (0.71)	59.24c (0.85)	2.69	.046	1 < 2, 3, 4	2 = 3 = 4
Withdrawn-depressed	56.92a,b,c (0.37)	58.35a (0.59)	58.54b (0.72)	59.99c (0.86)	4.59	.003	1 < 2, 3, 4	2 = 3 = 4
Somatic complaints	57.62a,b (0.36)	59.77a (0.58)	58.79 (0.71)	60.81b (0.85)	5.95	.001	1 < 2, 4	1 = 3; 2 = 3 = 4
Internalizing problems	57.04a,b,c (0.36)	59.67a (0.58)	59.42b (0.71)	60.58c (0.85)	5.99	<.001	1 < 2, 3, 4	2 = 3 = 4
Rule-breaking behaviors	55.07a,b (0.44)	56.95a (0.70)	56.23 (0.87)	56.73b (1.04)	5.01	.002	1 < 2, 4	1 = 3; 2 = 3 = 4
Aggressive behaviors	53.94a,b,c (0.27)	55.39a (0.43)	55.38b (0.53)	55.83c (0.63)	5.00	.002	1 < 2, 3, 4	2 = 3 = 4
Externalizing behaviors	51.29a,b,c (0.44)	54.09a (0.70)	53.74b (0.86)	54.10c (1.03)	5.59	.001	1 < 2, 3, 4	2 = 3 = 4
Attention problems	56.24a,b,c (0.30)	57.83a (0.49)	58.11b (0.60)	58.57c (0.72)	5.44	.001	1 < 2, 3, 4	2 = 3 = 4
Total problems	52.46a,b,c (0.39)	55.15a (0.63)	54.96b (0.78)	55.55c (0.93)	7.07	< .001	1 < 2, 3, 4	2 = 3 = 4

Note. Estimated mean values with the same letter superscript in the same row are significantly different (P < .05). Analyses adjusted for young adults’ sex, age, mother education, maternal IQ, family SES, and family stress.

^aThe ‘increasing’ trajectory reflects increasing depressive symptoms at child’s adolescence.

^bThe ‘decreasing’ trajectory reflects decreasing depressive symptoms at child’s adolescence.

Children’s Cognitive Functioning and Achievement at Adolescence by Maternal Depression Trajectory

Results of contrasts of adolescents’ cognitive test scores (Table 5) showed that, for all but the WISC matrix reasoning subtest, adolescents of mothers with chronically severe depression (class 4) had lower cognitive and achievement test scores than adolescents of mothers with infrequent symptoms (class 1). No other group differences were found.

Table 5.

Children’s Cognitive and Achievement Test Scores at Adolescence by Mothers’ Depression Trajectory

	Class 1 Infrequent (n = 474)	Class 2 Increasinga (n = 176)	Class 3 Decreasingb (n = 124)	Class 4 Chronic severe (n = 95)
	M (SE)	M (SE)	M (SE)	M (SE)	F (df)	P	Class contrasts	Equivalent classes
WISC – Language Similarities	8.56 (0.10)	8.34 (0.16)	8.31 (0.19)	7.87 (0.21)	3.05 (3,869)	.03	1 > 4	1 = 2 = 3 2, 3 = 4
WISC – Matrix Reasoning	7.62 (0.11)	7.33 (0.18)	7.63 (0.21)	7.32 (0.24)	0.98 (3,869)	ns
Reading – Vocabulary	22.18 (0.33)	21.48 (0.54)	21.23 (0.65)	19.57 (0.75)	3.48 (3,829)	.02	1 > 4	1 = 2 = 3 2, 3 = 4
Reading – Comprehension	12.05 (0.22)	11.47 (0.36)	11.64 (0.43)	10.34 (0.50)	3.34 (3,845)	.02	1 > 4	1 = 2 = 3 2, 3 = 4
WRAT – Mathematics	83.30 (0.44)	81.26 (0.72)	81.90 (0.86)	80.35 (0.99)	3.63 (3,867)	.02	1 > 4	1 = 2 = 3 2, 3 = 4

Note. Analyses adjusted for youth sex, youth age at testing, and mothers’ IQ, education, family stress and family SES. Adjusted scores are shown. ns = not statistically significant. WISC = The Wechsler Intelligence Scale for Children (3 ed.) Reading = The Spanish Reading Test (El Examen de Lectura en Español). WRAT = Wide Range Achievement Test.

^aThe ‘increasing’ trajectory reflects increasing depressive symptoms at child’s adolescence.

^bThe ‘decreasing’ trajectory reflects decreasing depressive symptoms at child’s adolescence.

Discussion

We identified four distinct trajectories of mothers’ depression. While over half of Chilean mothers in this sample experienced relatively infrequent depression, 11% - or roughly 1 in 10 mothers had severe depression till their child reached adolescence. For these mothers, stresses related to unemployment or underemployment, low education, patriarchal (male-controlled) gender roles, and living in small crowded homes with extended family may have contributed to their depression. Having many children at young ages, somewhat typical in developing countries, also could be a contributing factor (Shidhaye & Giri, 2014).

We also found that the various patterns of maternal depressive symptoms related to their children’s long-term adjustment. Results showed that youth of mothers with chronically severe depression reported significantly more internalizing symptoms themselves (anxiety, depression, withdrawal) at adolescence than youth of mothers with infrequent depressive symptoms. This pattern is consistent with other research that shows associations between mothers’ persistent high depression and their child’s behavioral and emotional symptoms (e.g., Cents et al., 2013; Kingston et al., 2018; Matijaserich et al., 2016; van der Waerden et al., 2015). However, these studies used maternal ratings of their child when children were age 6 years or younger. Current findings are based on youths’ own reports of their functioning and show that such relations to their mothers’ depression persist into adolescence.

It is noteworthy that adolescents whose mothers showed decreasing depressive symptoms (class 3) reported equivalent levels of anxiety, depression, withdrawal and somatic complaints as adolescents whose mothers had infrequent depression. This suggests a relatively positive outlook for adolescents whose mothers had significant depressive symptoms up to the child’s middle childhood but showed stark improvement during the child’s adolescence. Thus, alleviating mothers’ depressive symptoms even as late as when their children approach adolescence may help prevent socioemotional problems in their children at adolescence.

When examining adolescents’ externalizing and attention problems, youth whose mothers showed infrequent depression reported significantly lower levels of aggressive, rule-breaking, and attention problems than all other groups. This pattern persisted into young adulthood, with young adults whose mothers showed infrequent depression reporting significantly lower levels of anxiety, depression, aggression, attention problems and externalizing behaviors than all other groups. Thus, exposure to maternal depression of any extent, duration or time period during childhood appears to portend to higher levels of behavior and attention problems at both adolescence and adulthood and to higher levels of internalizing and externalizing problems at adulthood. This finding illustrates the significance and long-term ramifications of mothers’ depressed mood for their children’s mental and psychosocial health.

For results pertaining to cognitive and achievement testing, we found that adolescents of mothers with chronic severe depressive symptoms had significantly lower language, vocabulary, comprehension, and mathematic abilities than adolescents of mothers with infrequent depressive symptoms. Previous studies have found similar trajectory group differences in children’s cognitive abilities as a function of mothers’ depression at school entry (Campbell et al., 2007) and 3 years of age (Park et al., 2018). The current results show that lower cognitive scores among children of chronically depressed mothers persist beyond childhood and pertain to a broad range of cognitive abilities. Effects on children’s cognitive skills have been discussed as stemming from depressed mothers’ less responsive and engaging interactions with their child (Field, 2010; Wu et a., 2019). Other research shows that depressed mothers are less affectionate, playful, and verbal than non-depressed mothers (Goodman, 2007; Lovejoy et al., 2000), all attributes likely to affect children’s language and verbal skills.

Given evidence that maternal depression patterns relate to their child’s adjustment, it is important to consider what factors drive differences in maternal depression across time. Some research suggests that mothers’ adjustment varies by the developmental stage of their child (Luthar & Cicola, 2016). For example, while some mothers feel emotionally exhausted caring from a needy infant, other mothers find the increasing independence and rule-breaking behaviors of their adolescent children especially disheartening (Wu et al., 2019). Still other mothers may find that the pubertal changes in their child during early adolescence reflects a growing up and growing away of their child (Luthar & Cicola, 2016). Additionally, cultural factors play a role in maternal depression generally, such as culture-specific factors in level of support services available, stigma of mental health issues, socioeconomic factors, stress, and gender roles (Halbreich & Karkun, 2006). Further research that clarifies mothers’ mental health during the stages of her child’s life would help identify when and which mothers would benefit from mental health resources.

Several study features should be considered when interpreting the findings. We note first that neither mothers’ depression nor youths’ adjustment problems were assessed by clinicians, but rather by self-report on questionnaires. It is possible that symptoms may have been underreported. In addition, none of the average T scores at adolescence or young adulthood exceeded 65, the cutoff corresponding to a borderline clinical range. Thus, for the most part, the psychosocial functioning of youth in the current sample was in the normal range. However, some youth may indeed have been experiencing serious psychosocial problems. Analysis of individuals with T scores > 65 would be helpful here and an area for further study. It should also be recognized that this Chilean sample may not be representative of mothers or children of other cultural backgrounds or nationalities.

Mothers included in analyses had lower depression scores at their child’s infancy than mothers not studied. Thus, mothers’ initial depression levels may be positively skewed and this may have affected class membership. Adolescents included in the analyses were more likely to be from higher SES families and had mothers with slightly higher IQs. We adjusted for these and other factors, but the relatively more advantaged backgrounds of current participants raise the possibility that our results underestimate actual effects. Additionally, mothers reported depressive symptoms only within the last week, which captures a fairly constrained timeframe. We did not have information about whether mothers sought treatment for or were taking medication for their depressive symptoms, which could have affected across-time changes. In addition, assessment of youths’ psychosocial problems at adolescence occurred over a relatively wide age range, or from ages 12 to 17 years. Youths’ age was controlled in all analyses and did not differ across the maternal trajectory groups at adolescence (mean age range: 14.5 to 14.9) or young adulthood (mean age range: 20.5 to 20.6). However, youths’ relatively wide age range at the adolescent psychosocial assessment should be considered in interpreting study findings. Fathers’ depression was not assessed in this study, which is a known contributor to both mothers’ and children’s emotional difficulties (Cheung & Theule, 2019; Narayanan & Naerde, 2016). Although we statistically controlled for several mother, child and family characteristics in analyses in attempts to adjust for these factors, there are many additional factors that could affect maternal depression or children’s adjustment that were unmeasured and remain unaccounted for.

This study also has several notable strengths, including the large sample of mothers and youth followed longitudinally for over two decades. This allowed for a more thorough understanding of how across-time patterns of mothers’ depression affect their children’s long-term functioning. Additionally, a high percentage of mothers had significant depressive symptoms. Approximately half of mothers had CES-D scores ≥ 16 at any time during the study (reflecting moderate-to-severe depressive symptomatology) (Radloff, 1977), compared to less than 10% of mothers who had significant depressive symptoms in many previous studies. High rates of maternal depression have been found in the U.S. and in many countries, including developing countries (Ertel et al., 2011; Gelaye et al., 2016; Wolf et al., 2002). Thus, our findings may be relevant to many mothers around the world. Additionally, youth provided their own assessment of their functioning, which avoids the shared-reporter bias created by mothers providing their and their child’s mental health assessment (Lohaus et al., 2019).

Conclusions

Study findings suggest that early detection and continued monitoring of mothers’ depressive symptoms throughout their children’s development would help prevent long-term socioemotional problems in children and promote optimal child cognitive functioning. Universal screening of mothers during pregnancy would be ideal, particularly in low-income countries where maternal depression is prevalent and a significant threat to the health and well-being of mothers and children (Moussavi et al., 2007; Wachs et al., 2009). Screening of maternal depressive symptoms during all child pediatric visits might also improve detection and prevention. Chile is a developed middle-income country with access to national health care and health insurance (Gitlin & Fuentes, 2012). However, like all countries with national health care, access to high-quality professionals is often limited and delayed. Access to good quality mental health care is also often lacking in the U.S. Providing low-cost, high-quality, nonstigmatized, and accessible mental health care to mothers would aid significantly in the prevention and treatment of what is now widely recognized as a worldwide public health issue.

Highlights.

• Eleven percent – or 1 in 10 - Chilean mothers experienced frequent depression throughout their child’s childhood.

• Adolescents whose mothers had chronic severe depression had more frequent internalizing problems than other groups.

• Exposure to maternal depression of any severity or duration during childhood portended worse adjustment at adulthood.

• Chronic severe maternal depression related to adolescents’ poorer verbal and math abilities.