Maternal trauma history, maternal mental health, and child behavior: A prospective study of mother-child dyads in Lima, Peru

Abstract

This study investigates the association between maternal trauma history (lifetime exposure to physical and sexual violence), maternal mental health and child behavior. Pregnant women attending the Instituto Nacional Materno Perinatal (INMP) in Lima, Peru, were recruited to join the study beginning in February 2012. Maternal trauma history and prenatal mental health were assessed at 16-weeks gestation and mothers completed the Child Behavior Checklist when children were four years old. We used causal mediation analysis to examine the association between maternal trauma history and child behavior. This study population included 631 mother-child dyads; 74% of mothers reported one or more lifetime episodes of intimate partner violence (IPV). We found that 32% of the association between maternal lifetime exposure to IPV and child internalizing behavior was mediated by prenatal anxiety. Our findings suggest that prevention and treatment of perinatal IPV and mental health disorders should be prioritized.

Introduction

Sexual and physical violence against women and girls is a major public health problem worldwide. The World Health Organization (WHO) estimates that almost one in three women globally have experienced physical or sexual violence from an intimate partner.¹ Exposure to physical and sexual violence has been linked to a variety of short- and long-term adverse health outcomes, including depression,² post-traumatic stress disorder (PTSD),³ suicidal behavior,⁴ poverty,⁵ and chronic inflammation. ⁶ Childhood abuse has also been shown to increase the risk of being a victim of intimate partner violence as an adult. ⁷ Abuse not only affects the individual, but it can also have an intergenerational effect. Maternal exposure to violence during childhood ^{7, 8,9} and prenatally¹⁰ have been independently linked to increased risk of mental health disorders among their children, including social-emotional and behavioral problems, through what is known as the intergenerational transmission of trauma (ITT).¹¹ Less is known about the mechanisms underlying this association.

Various conceptual models have been proposed to explain the mechanisms underlying the association between maternal mental health and child behavior, but further evidence is needed to support them.^12,13,14 The developmental origin of health and disease (DOHaD) theory suggests that in-utero exposures can alter the risk of conditions later in life and across generations. ^10,15 Maternal stress during pregnancy may have physiological effects on the fetus including increased activation of the hypothalamic-pituitary-adrenal (HPA) axis¹⁶ and other stress processes, leading, for example, to dysregulated cortisol output throughout the life course. ¹¹ In support of this theory, a growing body of evidence has found an association between maternal depression in pregnancy and child behavior, independently of postnatal depression. ^{14, 17, 18} Maternal postpartum mental health has been identified as a proximal process that affects child behavior and socioemotional development, potentially through maternal responsiveness and mother-child interactions.^19,20

Women have an elevated risk of experiencing mental health disorders during and after pregnancy. The most widely studied of these is prenatal depression, which affects approximately 17% of women in high-income countries ²¹ and one in four women in low-and-middle income countries (LMIC). ²² Although screening for perinatal anxiety is less common, it also affects an estimated one in four women in LMIC. ²³ There is less evidence about the prevalence of perinatal PTSD), particularly in LMIC, but evidence suggests it may be underdiagnosed.²⁴

Child social-emotional and behavior problems, including internalizing behavior (such as withdrawal and anxiety) and externalizing behavior (such as aggression and impulsivity) have been linked to difficulties with academic and social development^25,26 as well as an increased risk of mental health disorders later in life.²⁷ Further longitudinal evidence is needed on the risk factors and mechanisms underlying child behavior problems to develop effective prevention strategies. While maternal mental health plays a key role in both DOHaD and bioecological models, only a few studies conducted in high-income countries have examined maternal depression as a potential mediator of the association between maternal history of abuse and child behavior ^28,29,30 and none to our knowledge have looked at PTSD. Additionally, while depression and anxiety are frequently comorbid and are more likely to occur together during the perinatal period,¹⁸ few studies have looked at their relative contributions to child behavior.

The present study explored the direct and indirect relationships between maternal exposure to violence (hereafter referred to as “maternal trauma history”), maternal mental health, and child internalizing and externalizing behavior in a cohort of mother and child dyads in Lima, Peru. Peru is an upper-middle-income, post-conflict country with one of the highest levels of violence against women worldwide.³¹ We assessed (1) the association between maternal trauma history and perinatal (prenatal and when children were four years old) maternal mental health; (2) the association between maternal trauma history and child behavior; (3) the association between perinatal maternal mental health and child behavior; and (4) whether prenatal maternal mental health mediates the relationship between maternal trauma history and child externalizing and internalizing behavior.

Methods

Procedure

The present study uses data from the ongoing Pregnancy Outcomes, Maternal and Infant Study (PrOMIS) longitudinal cohort study, which looks at maternal behavioral and social risk factors for pregnancy and child outcomes. The study setting and data collection process has been described previously.^3,4,32 Pregnant women attending the Instituto Nacional Materno Perinatal (INMP) in Lima, Peru, were recruited to join the study beginning in February 2012. The INMP is a primary referral hospital overseen by the Peruvian Ministry of Health, which provides maternal and perinatal care for low-income women with public insurance. Pregnant women between 18 and 49 years of age, at 16 weeks gestation or less, who spoke and understood Spanish were eligible to participate. Exclusion criteria included a twin pregnancy, known fetal malformation, maternal history of intellectual disability, chronic hypertension, diabetes mellitus, sepsis or renal failure. Enrolled participants were interviewed by trained research personnel in a private space at the INMP using structured questionnaires. This study uses data from two-time points: the first interview, which took place at approximately 16 weeks gestation, and a follow–up interview when children were, on average, four years old. Interviewers were Peruvian women employed as social workers at INMP. All procedures used in the study were approved by the INMP and Harvard T. H. Chan School of Public Health Institutional Review Boards and all mothers provided informed consent for their own and their child’s participation in the study.

Participants

The population of this study included women who were enrolled in the PrOMIS cohort between February 2012 and November 2015. A total of 5,440 women were interviewed during early pregnancy, and 4,472 had a live birth and 31 stillbirths. A follow-up pilot study on a subset of the original cohort and their children started in December 2017. The first wave of participants who were initially recruited were contacted to participate in the follow-up study; 631 mother-child dyads were successfully enrolled. Women who completed follow–up did not differ from the initial cohort in terms of sociodemographic characteristics but were more likely to have reported one or more instances of childhood abuse (74.4% vs. 58.64%) as compared to those who did not complete follow-up.

2.3. Measures

Maternal Childhood physical and sexual abuse

Maternal experience of childhood abuse was assessed prenatally using the Childhood Physical and Sexual Abuse Questionnaire.³³ This eight-item instrument is based on the Centers for Disease Control and Prevention (CDC) Adverse Childhood Experiences Study. Questions about physical abuse asked whether before the age of 18 years, an older person had hit, kicked, pushed, or beat them and/or threatened their life. Questions about sexual abuse asked whether, before the age of 18 years, an older person had touched them in a sexual way, they were made to touch someone else in a sexual way, or someone attempted or completed intercourse with them. The childhood abuse variable was created by summing the number of events and creating a binary variable for 0 events or ≥1 event This questionnaire has been used previously in a Peruvian population. ^4,32,34

Maternal Exposure to Intimate Partner Violence

Mothers answered questions during the prenatal interview about physical and sexual acts of violence committed against them by a current or former intimate partner during their lifetime. Questions were adapted from the Demographic and Health Survey Domestic Violence Module ³⁵ and the WHO Multi-Country Study on Violence Against Women.³⁶ The questionnaire included questions about controlling behavior (7 items), emotional abuse (4 items), physical violence (6 items), and sexual violence (3 items). A binary lifetime intimate partner violence (IPV) variable was created for 0 or ≥ 1 items. Interviewers were trained to assess if women were in physically dangerous situations and/or needing immediate counseling and referred them to local psychiatrists, psychologists, and women’s organizations as needed.

Maternal Depression

Symptoms consistent with maternal depression (hereafter referred to as depression) was measured at the prenatal and follow-up interviews using the Patient Health Questionnaire (PHQ-9).³⁷ The PHQ-9 uses a four-point scale: “not at all,” “several days,” “more than half the days,” or “nearly every day,” to assess depressive symptoms in the last 14 days, with a total score ranging from 0 to 27. A Rasch item response theory (IRT) analysis of the Spanish-language version of the PHQ-9 in the current study population demonstrated that it had the properties of an effective screening instrument³⁸ with a Cronbach’s alpha of 0.81.³⁹ We classified women with scores ≥10 as having symptoms consistent with depression, a widely used cutoff which has been shown to have a specificity of 0.80 and a sensitivity of 0,92 in diagnosing possible major depressive disorder.⁴⁰

Maternal Generalized Anxiety Disorder

Symptoms consistent with maternal generalized anxiety disorder (GAD) (hereafter referred to as anxiety) were assessed during the prenatal interview using the GAD-7, a 7-item questionnaire that uses a four-point scale “not at all,” “several days,” “more than half the days,” and “nearly every day,” to assess the frequency of anxiety symptoms.⁴¹ We observed a Cronbach’s alpha of 0.88 in our study sample.⁴² Women with scores ≥ 7 were classified as having symptoms consistent with GAD. A validation study conducted among pregnant Peruvian women found that cutoff of 7 maximized the Youden Index and yielded a sensitivity of 73.3 and a specificity of 67.3.⁴²

Maternal Post-Traumatic Stress Disorder

Maternal symptoms consistent with PTSD) were assessed at both the first and second interviews using the PTSD Checklist-civilian (PCL-C). The PCL-C is a widely used self-reported questionnaire that consists of 17 items corresponding to the DSM IV criteria for PTSD.⁴³ For each question, respondents are asked to assess how much they have been bothered by that symptom in the last month: 1 (not at all), 2 (a little bit), 3 (moderately), 4 (quite a bit), and 5 (extremely). A validation study of the Spanish language version of the PCL-C conducted in a subset of this study population during pregnancy found that it had good reliability (Cronbach’s alpha =0.90)., criterion validity, and factorial validity.⁴⁴ We define PTSD as a PCL-C score ≥26, a cutoff which had 86% sensitivity and 63% specificity in diagnosing PTSD in a Peruvian population.⁴⁴

Child Behavior

Mothers reported child behavior during follow–up using the Child Behavior Checklist (CBCL) for children ages 1.5–5 years.⁴⁵ The CBCL/1.5–5 is a 100-item instrument that measures caregivers’ perception of a child’s emotional, behavioral, and social problems in the last two months. Seven syndrome scales were identified through exploratory and confirmatory factor analysis^,45 representing common problems for this age group: Emotionally Reactive (9 items), Anxious/Depressed (8 items), Somatic Complaints (11 items), Withdrawn (8 items), Sleep Problems (7 items), Attention Problems (5 items), and Aggressive (19 items). The Internalizing Problems variable is a combination of the Emotionally Reactive, Anxious/Depressed, Somatic Complaints, and Withdrawn items. The Externalizing Problems score is a combination of the Attention Problems and Aggressive items (Sleep Problems are treated as a separate problem). This analysis uses T scores derived from this normative sample for the CBCL/1.5–5.⁴⁵ A validation study found that this seven-factor model had an acceptable fit in a Peruvian sample.⁴⁶ We used Achenbach’s (1991) cutoff of a T-score ≥ 70 as the threshold for clinically significant problems at the syndrome scale level and ≥ 64 for the total internalizing and externalizing scores.⁴⁷

Analytic Plan

For each of the 7 child behavior syndrome scales, we calculated the mean scores and percentage of children who scored above the clinical threshold of 70. We used log-binomial models which produced relative risk estimates (RR), to examine the association between maternal history of exposure to violence (exposure to lifetime IPV and childhood abuse) and clinically significant child internalizing and externalizing behavior problems. Next, we examined the association between maternal mental health and child behavior. We first looked at maternal prenatal depression, anxiety, and PTSD as exposures, both independently and in combination. For both sustained PTSD and depression, we divided women into four categories: (1) Women who scored above the clinical threshold at both baseline and follow-up (“sustained”); (2) at baseline only (“prenatal”); (3) at follow-up only (“postnatal”); and (4) at neither time-point. We then examined the relationship between sustained maternal mental health and child behavior problems to explore the relative contributions of depression and PTSD at depression and follow-up on child behavior. We assessed the risk of child behavior problems for each category, using the “neither” category as the reference group.

We used log-binomial models to examine the association between maternal history of trauma (exposure to lifetime IPV and childhood abuse) and maternal prenatal mental health disorders. Additionally, we used multinomial (polytomous) regression for the sustained mental health analyses to simultaneously explore the association between maternal trauma history and PTSD at different time points. We reported the risk of prenatal, postnatal, and sustained PTSD relative to the “no PTSD at either time” group.

We used causal mediation analysis to assess the proportion of the association between maternal trauma history (maternal lifetime IPV and childhood abuse) and child behavior problems that is explained by maternal prenatal depression, anxiety, and PTSD, as well as sustained depression and PTSD. We used the R package CMAverse to estimate the total effect, indirect (“mediated”) effect, direct (“unmediated”) effect, and proportion mediated.⁴⁸ We used the g-formula approach as it is doubly robust and accommodates the presence of mediator-outcome confounders that are associated with the exposure.⁴⁹ This approach uses imputation-based estimation, in which, the number of draws is restricted to the analytic sample size. As this can introduce some Monte Carlo variability, we used bootstrap resampling (n = 1000 to obtain confidence intervals.”.)

All multivariate models were adjusted for maternal age, parity (nulliparous at baseline vs. multiparous), difficulty paying for basic needs such as food (hard vs. not very hard), and marital status (married/living with partner vs. other). Covariates were determined a priori based on a review of the literature. ^11,22 Missingness of exposure, mediator, and covariate variables ranged from 0.2% to 1.1%. R 4.2.0 was used to perform all analyses.⁵⁰

Results

Study population sociodemographic characteristics

This study population included 631 mother-child dyads who completed the Child Behavior Checklist when the children were approximately four years of age. Select maternal sociodemographic characteristics at their prenatal interview are presented in Table 1. The mean age of pregnant women at the prenatal interview was 29.1 years, ranging from 18 to 44 years. Nearly half of the participants had 12 years of education or more (46.4%), and 46.9% were employed. Most participants were married or cohabitating with a partner (81.6%) and described their race/ethnicity as mestizo (77.2%). For 46.1% of participants, this baby was their first child. Approximately 47.7% of participants had difficulty accessing basic food, and 53.2% had difficulty paying for medical care. Nearly three-quarters of participants reported experiencing one or more episodes of childhood physical or sexual abuse (74.2%) and 37.8% reported one or more lifetime episodes of IPV. Prenatal depressive symptoms were reported by 24.5% of participants, prenatal anxiety by 30%, and prenatal symptoms PTSD symptoms by 37.2%. Approximately 13.7% of participants scored above the clinical threshold for all three. At the follow-up interview depressive symptoms were reported by 28.7% of participants and PTSD symptoms were reported by 49.0% of participants (anxiety was not measured at this interview). Approximately 9.0% of participants reported depressive symptoms at both interviews and 25.2% reported symptoms of PTSD at both interviews.

Table 1:

Characteristics of pregnant women at prenatal interview (N=631)

Characteristics	N (%)	Missing N (%)
Socio-demographics
Age		1 (0.2)
18-19	24 (3.8)
20-29	315 (49.9)
30-34	163 (25.8)
≥ 35	128 (20.3)
Mean (SD)	29.1 (6.1)
Education (years)		2 (0.3)
≤ 6	25 (4.0)
7-12	311 (49.3)
>12	293 (46.4)
Mestizo ethnicity	487 (77.2)	2 (0.3)
Married/living with a partner	515 (81.6)	3 (0.5)
Employed	296 (46.9)	2 (0.3)
Access to basic foods		2 (0.3)
Hard	301 (47.7)
Not very hard	328 (52.0)
Difficulty in paying for medical care		1 (0.2)
Hard	335 (53.2)
Not very hard	295 (46.8)
Nulliparous	291 (46.1)	1 (0.2)
Maternal history of trauma
Childhood abuse	467 (74.2)	2 (0.2)
Lifetime intimate partner violence	236 (37.8)	6 (1.0)
Intimate partner violence in the last 12 months	96 (15.2)	7 (1.1)
Overall trauma exposure		6 (1.0)
Both maternal childhood abuse and lifetime intimate partner violence	196 (31.1)
Lifetime intimate partner violence only	40 (6.3)
Childhood abuse only	268 (42.5)
Neither	121 (19.2)
Prenatal mental health
Prenatal depression (PHQ9≥10)	154 (24.5)	3 (0.5)
Prenatal anxiety (GAD-7≥ 7)	201 (32.2)	7 (1.1)
Prenatal PTSD (PCL-C≥ 26)	234 (37.2)	3 (0.5)
Prenatal depression, anxiety, and PTSD	86 (13.7)	5 (0.8)
Mental health at follow-up
Postnatal depression (PHQ9≥10)	180 (28.7)	3 (0.5)
Postnatal PTSD (PCL-C≥26)	308 (49.0)	2 (0.2)
Sustained depression	56 (9.0)	6 (1.0)
Sustained PTSD	158 (25.2)	5 (0.8)

Prevalence of child internalizing and externalizing problems

The prevalence of child behavior problems is summarized in Supplementary Table 1. Almost a third of children scored above the clinical threshold for internalizing problems (30.6%). The most common internalizing syndrome was anxious-depressed behavior (16.3%), followed by somatic complaints (13.8%), withdrawn behavior (11.6%), and emotionally reactive behavior (7.3%). The prevalence of externalizing behavior problems was 17.3%, with 4.6% of children scoring within the clinical range for aggressive behavior and 7.0% for attention problems.

Exposure-mediator associations (maternal trauma history and maternal mental health)

Table 2 reports the associations between maternal trauma history and maternal mental health at the first and follow-up interviews. Childhood abuse was associated with prenatal depression, anxiety, and PTSD, as was lifetime IPV (Adjusted Relative Risk (aRR): 2.18, 95% CI:1.44, 3.29), anxiety (aRR: 1.97, 95% CI: 1.40, 2.78), and PTSD (aRR: 2.02, 95% CI: 1.47, 2.77). Lifetime exposure to IPV was also associated with prenatal depression (aRR: 1.92, 95% CI: 1.46, 2.53), prenatal anxiety (aRR: 1.92, 95% CI: 1.52, 2.42), and prenatal PTSD (aRR: 1.88, 95% CI: 1.53, 2.31). Supplementary Table 4 shows the association between maternal trauma history and sustained maternal depression, while Supplementary Table 5 shows the association between trauma history and sustained PTSD. In multinomial regression models adjusting for maternal age, parity, difficulty paying for basic needs, and marital status, both maternal history of childhood abuse and lifetime IPV were independently associated with sustained PTSD. Compared to women who experienced neither childhood abuse nor lifetime IPV, women who experienced both had 9.31 times the risk of having sustained depression (95% CI: 3.07, 28.19) and over twelve times the risk of having sustained PTSD relative to no PTSD (Adjusted Relative Risk (aRR): 12.2; 95% CI: 5.63, 26.45).

Table 2: Associations between exposure and mediators (maternal trauma history and prenatal mental health) (N = 631).

	Prenatal depression (N=154)		Prenatal anxiety (N=201)		Prenatal PTSD (N=234)
Trauma history	N (%)	aRR1 (95% CI)	N (%)	aRR1 (95% CI)	N (%)	aRR1 (95% CI)
Maternal childhood abuse
Yes	132 (85.7)	2.18 (1.44, 3.29)	171 (85.1)	1.97 (1.40, 2.78)	200 (85.5)	2.02 (1.47, 2.77)
No	22 (14.3)	Ref.	30 (14.9)	Ref.	34 (14.5)	Ref.
Maternal lifetime intimate partner violence
Yes	81 (53.3)	1.92 (1.46, 2.53)	106 (53.3)	1.92 (1.52, 2.42)	124 (53.7)	1.88 (1.53, 2.31)
No	71 (46.7)	Ref.	93 (46.7)	Ref.	107 (46.3)	Ref.
Maternal intimate partner violence in the last 12 months
Yes	28 (18.3)	1.19 (0.84, 1.69)	48 (24.0)	1.77 (1.39, 2.24)	52 (22.6)	1.59 (1.28, 1.97)
No	125 (81.7)	Ref.	152 (76/0)	Ref.	178 (77.4)	Ref.
Overall trauma exposure
Both maternal childhood abuse and lifetime intimate partner violence	74 (48.7)	3.26 (1.97, 5.40)	92 (46.2)	3.67 (2.26, 5.94)	110 (47.6)	3.38 (2.22, 5.17)
Lifetime intimate partner violence only	7 (4.6)	1.38 (0.60, 3.17)	14 (7.0)	2.67 (1.43, 4.97)	14 (6.1)	2.08 (1.16, 3.72)
Childhood abuse only	56 (36.8)	1.1.76 (1.04, 2.98)	77 (38.7)	2.19 (1.33, 3.58)	87 (37.7)	11.98 (1.28, 3.07)
Neither	15 (9.9)	Ref.	16 (8.0)	Ref.	20 (8.7)	Ref.

¹Adjusted for maternal age, parity, difficulty to pay for the basics, married/living with a partner.

Exposure-outcome association (maternal trauma history and child behavior)

Supplementary Table 3 presents the association between maternal trauma history and child behavior problems. We did not observe a statistically significant association between maternal experience of child abuse and internalizing (aRR: 1.08, 95% CI: 0.82, 1.42) or externalizing behavior (aRR: 1.28, 95% CI: 0.83, 1.95). Women who reported lifetime exposure to intimate partner violence in the last 12 months were more likely to report both child internalizing behavior problems (aRR: 1.68, 95% CI: 1.31, 2.16) and child externalizing problems (aRR: 1.56, 95% CI: 1.04, 2.34). Women who reported lifetime exposure to IPV were more likely to report exposure to child internalizing behavior (aRR: 1.46, 95% CI: 1.16, 1.85) but not child externalizing behavior (aRR: 1.23 (95% CI: 0.86, 1.75).

Mediator-outcome association (maternal mental health and child behavior)

The association between maternal mental health and child internalizing problems is shown in Table 3.

Table 3: Associations between mediators and outcomes (maternal mental health and child internalizing and externalizing problems) (N = 631).

		CBCL Internalizing Symptoms T-score ≥ 64 (N=193)			CBCL Externalizing Symptoms T-score ≥ 64 (N=109)
Maternal mental health	N (%)	RR (95% CI)	aRR (95% CI)	N (%)	RR (95% CI)	aRR (95% CI)
Prenatal depression
Yes	48 (24.9)	1.02 (0.78, 1.34)	0.99 (0.76, 1.30)	28 (25.9)	1.08 (0.73, 1.59)	1.08 (0.73, 1.60)
No	145 (75.1)	Ref.	Ref.	80 (74.1)	Ref.	Ref
Prenatal generalized anxiety (GA)
Yes	81 (42.2)	1.54 (1.22, 1.94)	1.49 (1.18, 1.87)	45 (41.7)	1.50 (1.07, 2.12)	1.48 (1.05, 2.09)
No	111 (57.8)	Ref.	Ref.	63 (58.3)	Ref.	Ref
Prenatal PTSD
Yes	84 (43.5)	1.30 (1.03, 1.64)	1.26 (1.00, 1.59)	56 (51.9)	1.81 (1.29, 2.55)	1.76 (1.25, 2.49)
No	109 (56.5)	Ref.	Ref.	52 (48.1)	Ref.	Ref
Prenatal depression and anxiety
Both depression and GA	39 (20.3)	1.37 (1.02, 1.85)	1.30 (0.96, 1.75)	24 (22.2)	1.46 (0.95, 2.22)	1.43 (0.94, 2.19)
GA only	42 (21.9)	1.61 (1.22, 2.14)	1.62 (1.23, 2.13)	21 (19.4)	1.39 (0.89, 2.17)	1.38 (0.88, 2.16)
Depression only	9 (4.7)	0.71 (0.38, 1.30)	0.73 (0.40, 1.34)	4 (3.7)	0.54 (0.21, 1.43)	0.56 (0.21, 1.49)
Neither	102 (53.1)	Ref.	Ref.	59 (54.6)	Ref.	Ref
Prenatal depression and PTSD
Both depression and PTSD	37 (19.2)	1.24 (0.91, 1.69)	1.18 (0.87, 1.61)	24 (22,2)	1.64 (1.06, 2.55)	1.63 (1.05, 2.52)
PTSD only	47 (24.4)	1.28 (0.97, 1.70)	1.26 (0.95, 1.67)	32 (29.6)	1.78 (1.20, 2.66)	1.71 (1.14, 2.56)
Depression only	11 (5.7)	0.79 (0.46, 1.37)	0.80 (0.46, 1.37)	4 (3.7)	0.59 (0.22, 1.56)	0.59 (0.22, 1.56)
Neither	98 (50.8)	Ref.	Ref.	48 (44.4)	Ref.	Ref
Prenatal depression, anxiety, and PTSD
All 3	32 (16.7)	1.41 (1.01, 1.97)	1.34 (0.96, 1.87)	21 (19.4)	1.74 (1.09, 2.78)	1.73 (1.09, 2.76)
1-2	81 (42.2)	1.28 (0.99, 1.66)	1.27 (0.99, 1.65)	45 (41.7)	1.34 (0.91, 1.97)	1.32 (0.90, 1.95)
None	79 (41.1)	Ref.	Ref.	42 (38.9)	Ref.	Ref
Sustained mental health symptoms
Depression
Sustained depression	29 (15.0)	2.51 (1.81, 3.48)	2.49 (1.81, 3.42)	19 (17.6)	3.38 (2.09, 5.48)	3.41 (2.10, 5.50)
Depression at follow-up only	73 (37.8)	2.88 (2.23, 3.70)	2.88 (2.23, 3.72)	45 (41.7)	3.65 (2.47, 5.39)	3.66 (2.43, 5.50)
Prenatal depression only	19 (9.8)	0.95 (0.60, 1.49)	0.93 (0.59, 1.46)	9 (8.3)	0.93 (0.46, 1.86)	0.94 (0.46, 1.88)
No depression at either time	72 (37.3)	Ref.	Ref.	35 (32.4)	Ref.	Ref
PTSD
Sustained PTSD	75 (38.9)	2.92 (2.10, 4.05)	2.95 (2.12, 4.10)	45 (41.7)	4.67 (2.70, 8.09)	4.83 (2.74, 8.53)
PTSD at follow-up only	69 (35.8)	2.87 (2.06, 4.00)	2.93 (2.10, 4.09)	37 (34.3)	4.10 (2.33, 7.21)	4.32 (2.42, 7.71)
Prenatal PTSD only	9 (4.7)	0.75 (0.38, 1.47)	0.71 (0.36, 1.40)	11 (!0.2)	2.44 (1.17, 5.08)	2.45 (1.16, 5.17)
No PTSD at either time	40 (20.7)	Ref.	Ref.	15 (13.9)	1.08 (0.73, 1.59)	1.08 (0.73, 1.60)

¹Adjusted for maternal age, parity, difficulty to pay for the basics, and married/living with a partner.

Depression

We did not observe a significant association between prenatal depression and child internalizing problems (aRR: 0.99, 95% CI: 0.76, 1.30) or externalizing problems (aRR: 1.08, 95% CI: 0.73, 1.60). However, women who had depression at follow-up only had an increased risk of child internalizing problems (aRR: 2.88; 95% CI: 2.23, 3.72), as did women who had both prenatal depression and depression at follow-up (“sustained depression”) (aRR: 2.49; 95% CI: 1.81, 3.42). Women who reported postnatal depression only were more than three times more likely to report child externalizing behavior problems above the clinical threshold (aRR: 3.66; 95% CI: 2.43, 5.50), compared to women with no depression at any time. We observed a similar effect size among women who had sustained depression (aRR: 3.41; 95% CI: 2.10, 5.54).

Anxiety

Women who experienced prenatal anxiety were significantly more likely to report both child internalizing problems (aRR: 1.49; 95% CI: 1.18, 1.87) and child externalizing problems (aRR: 1.48; 95% CI: 1.05, 2.09) above the clinical threshold

PTSD

Prenatal PTSD was associated with both internalizing (aRR: 1.26, 95% CI: 1.00, 1.59) and externalizing behavior (aRR: 1.76, 95% CI: 1.25, 2.49). In the multinomial sustained PTSD model, women who had prenatal PTSD only were not more likely to report child internalizing problems compared to women with no PTSD (aRR: 0.71, 95% CI: 0.36, 1.40), but both PTSD at follow-up only and sustained PTSD were significantly associated with internalizing problems (aRR: 2.93; 95% CI: 2.10, 4.09; aRR: 2.95; 95% CI: 2.12, 4.10, respectively). Women who had only prenatal PTSD were more likely to report child externalizing behavior problems compared to women with no PTSD at any time (aRR: 2.45; 95% CI: 1.16, 5.17), as were women with postnatal PTSD only (aRR: 4.32: 95% CI: 2.42, 7.71). Women with sustained PTSD were 4.83 times as likely to report child externalizing behavior problems (95% CI: 2.74, 8.53).

Mediation analysis

The results of the mediation analysis are reported in Table 4.

Table 4 – Mediation analysis for the association between maternal lifetime exposure to IPV and child behavior¹.

Panel A: Association with internalizing problems
	Lifetime IPV	IPV in the last 12 months	Childhood abuse
Mediators	aOR (95% CI)	aOR (95% CI)	aOR (95% CI)
Prenatal depression
Unmediated effect	1.74 (1.21, 2.52)	2.31 (1.48, 3.53)	1.09 (0.75, 1.68)
Mediated effect	0.98 (0.90, 1.09)	1.02 (0.94, 1.11)	1.00 (0.93, 1.07)
Total effect	1.72 (1.21, 2.45)	2.29 (1.47, 3.50)	1.08 (0.76, 1.66)
Percent mediated	N/A2	NA	N/A
Prenatal anxiety
Unmediated effect	1.48 (1.00, 2.15)	2,04 (1.27, 3.26)	0.97 (0.66, 1.47)
Mediated effect	1.15 (1.03, 1.37)	1.17 (0.98, 1.48)	1.08 (1.02, 1.22)
Total effect	1.70 (1.21, 2.43)	2.26 (1.43, 3.51)	1.05 (0.74, 1.63)
Percent mediated	32%	NA	N/A
Prenatal PTSD
Unmediated effect	1.58 (1.08, 2.34)	2.20 (1.43, 3.53)	1.00 (0.66, 1.51)
Mediated effect	1.09 (0.96, 1.27)	1.09 (0.90, 1.30)	1.08 (1.01, 1.23)
Total effect	1.72 (1.22, 2.46)	2.32 (1.50, 3.64)	1.07 (0.73, 1.663
Percent mediated	N/A	NA	N/A
Prenatal Depression, PTSD, Anxiety (joint effect)
Unmediated effect	1.46 (0.98, 2.18)	1.84 (1.08, 3.20)	0.92 (0.62, 1.53)
Mediated effect	1.21 (1.00, 1.40)	1.18 (0.94, 1.60)	1.12 (1.00, 1.29)
Total effect	1.77 (1.18, 2.49)	2.18 (1.43, 3.46)	1.08 (0.74, 160)
Percent mediated	40%	NA	NA
Panel B: Association with externalizing problems
	Lifetime IPV		Childhood abuse
Mediators	aOR (95% CI)	aOR (95% CI	aOR (95% CI
Prenatal depression
Unmediated effect	1.26 (0.79, 1.95)	1.76 (1.04, 2.99)	1.32 (0,80, 2.23)
Mediated effect	1.03 (090, 1.17)	1.07 (0.93, 1.25)	1.00 (0.92, 1.08)
Total effect	1.29 (0.83, 1.97)	1.72 (1.04, 2.89)	1.32 (0.80, 2.18)
Percent mediated	N/A	NA	N/A
Prenatal anxiety
Unmediated effect	1.09 (0.68, 1.74)	1.63 (0.99, 2.87)	1.22 (0.73, 2.14)
Mediated effect	1.17 (1.02, 1.45)	1.16 (0.98, 1.57)	1.11 (0.99, 1.24)
Total effect	1.28 (0.83, 2.04)	1.73 (1.06, 3.05)	1.35 (0.81, 2.34)
Percent mediated	N/A	NA	N/A
Prenatal PTSD
Unmediated effect	1.08 (0.63, 1.75)	1.54 (0.83, 2.61)	1.08 (0.63, 1.89)
Mediated effect	1.17 (1.02, 1.48)	1.16 (0.94, 1.48)	1.21 (1.07, 1.42)
Total effect	1.27 (0.79, 2.04)	1.80 (0.96, 3.01)	1.30 (0.79, 2.25)
Percent mediated	N/A	NA	N/A
Prenatal Depression, PTSD, Anxiety (joint effect)
Unmediated effect	1.16 (0.71, 1.84)	1.43 (0.80. 2.55)	1.25 (0.70, 2.47)
Mediated effect	1.30 (1.02, 1.63)	1.18 (0.85, 1.70)	1.23 (1.04, 1.46)
Total effect	1.32 (0.82, 1.96)	1.70 (1.00, 2.88)	1.39 (0.87, 2.40)
Percent mediated	NA	NA	NA

¹Adjusted for maternal age, parity, difficulty to pay for the basics, and married/living with a partner.

²Percent mediated only reported if both total association and mediated effect are statistically significant

Internalizing behavior

We found that 32% of the total effect between maternal lifetime IPV and child internalizing problems were mediated by maternal prenatal anxiety (adjusted odds ratio (aOR): 1.15, 95% CI: 1.03, 1.37). We did not observe an indirect effect between lifetime IPV and internalizing behavior through prenatal depression (aOR: 0.98, 95% CI: 0.90, 1.09) or prenatal PTSD (aOR: 1.09, 95% CI: 0.96, 1.27). We did not observe an indirect effect through prenatal depression (aOR: 1.02, 95% CI: 0.94, 1.11), anxiety (aOR: 1.17, 95% CI: 0.98, 1.48), or PTSD (aOR: 1.09, 95% CI: 0.90, 1.30) for the association between exposure to IPV in the previous 12 months and internalizing behavior, but we observed a statistically significant joint direct effect (aOR: 1.84, 95% CI: 1.08, 3.20). Although the joint total effect between exposure to child abuse and internalizing behavior was not statistically significant (aOR:1.08, 95% CI: 0.74, 160), we observed an indirect effect through prenatal anxiety (aOR: 1.08, 95% CI: 1.02, 1.22)) and PTSD (aOR:1.08, 95% CI: 1.01, 1.23) but not depression (aOR: 1.00 (95% CI: 0.93, 1.07).

Externalizing behavior

We did not observe a joint total effect between maternal history of IPV and child externalizing behavior (aOR: 1.32, 95% CI: 0.82, 1.96) but we observed a joint indirect effect of through prenatal mental health (aOR: 1.30, 95% CI: 1.02, 1.63), as well as indirect effects through anxiety (aOR: 1.17, 95% CI: 1.02, 1.45) and PTSD (aOR: 1.17, 95% CI: 1.02, 1.48) when examined in separate mediation models. Similarly, we did not observe a total joint effect between exposure to child abuse and child externalizing behavior (aOR: 1.39, 95% CI: 0.87, 2.40) but we observed a joint indirect effect through prenatal mental health (aOR: 1.23, 95% CI: 1.04, 1.46). In separate mediation models we observed an indirect effect through prenatal PTSD (aOR: 1.21, 95% CI: 1.07, 1.42), but not depression (aOR: 1.00, 95% CI: 0.92, 1.08) or maternal anxiety (aOR: 1.11, 95% CI: 0.99, 1.24).

Discussion

This study examined the relationships between maternal trauma history, maternal mental health, and child behavior problems. Maternal experience of IPV, but not child abuse, was associated with an increased risk of child internalizing and externalizing behavior. Prenatal anxiety and PTSD, but not prenatal depression, were associated with a higher risk of child internalizing and externalizing problems. Maternal PTSD and depression, assessed when children were approximately four years of age, were also associated with a higher risk of concurrent child internalizing and externalizing problems. Both maternal experience of childhood abuse and maternal lifetime experience of IPV increased the risk of prenatal depression, anxiety, and PTSD. Mothers who experienced both IPV and childhood abuse had over 9 times the risk of sustained depression and over 12 times the risk of sustained PTSD. Our findings suggest that more attention is needed to address maternal anxiety and PTSD, both during and after pregnancy, particularly among populations with high levels of exposure to trauma.

While we did not observe a total effect association between maternal experience of IPV and child externalizing problems, we observed indirect effects through prenatal anxiety and prenatal PTSD. We also observed indirect effects through prenatal anxiety and PTSD for the association between exposure to maternal childhood abuse and internalizing behavior, as well as an indirect effect through PTSD for the association between maternal childhood abuse and externalizing behavior, although the total effects were not statistically significant.. Only a few studies conducted in high-income countries have examined maternal mental health as a potential mediator of the association between maternal history of abuse and child behavior and none to our knowledge have looked at PTSD. One study conducted in Canada found that the association of maternal childhood physical abuse, but not childhood sexual abuse, with increased risk of internalizing problems in their children was mediated by maternal postnatal depressive symptoms.²⁸ Another study, also conducted in Canada, observed an indirect association between maternal adverse childhood events and child externalizing and internalizing problems through maternal depression and anxiety.²⁹

This study provides evidence from a largely low-income population in a middle-income country with a high prevalence of violence against women. In contrast to previous studies, we did not observe an association between maternal prenatal depression and child behavior problems.^14,17,18 Building on previous work on maternal postnatal psychopathology and child behavior,^7,8 we found that maternal postnatal depressive and PTSD symptoms are positively and independently associated with concurrent child behavior problems, underscoring the importance of current maternal mental health for child behavior. Previous research has shown that both maternal PTSD and depression are associated with impaired functioning across a variety of parenting domains, including inconsistent or reactive discipline and poor responsiveness to children’s needs.^51,52 Further research is needed to explore other potential mechanisms that underlie these associations.

This study is limited by several factors. First, since the maternal psychopathology at follow–up and the child behavior outcomes were collected concurrently when the children were approximately age four, we are unable to establish the direction of effect. It is possible that child behavioral problems exacerbated or caused maternal psychopathology symptoms. While the association between maternal PTSD symptoms during pregnancy and child externalizing behavior establishes temporality, further longitudinal research on this topic is needed with maternal postnatal psychopathology measures collected prior to child outcomes, including maternal pre-conception mental health. Given the robust literature on the relationship between maternal depression and impaired infant attachment,^53,54 depression data collected within the first six months postnatally would be useful. Second, all maternal variables included in the model were self-reported. Previous research has shown that maternal psychopathology may bias their reporting of child outcomes.^,5556 If mothers experiencing psychopathology symptoms were more likely to report child behavioral problems, we may be overestimating the magnitude of this association. However, even if a more objective measure of these children would not find a difference, it is worth noting that their mothers’ perception that they have more behavior problems could still increase their risk of experiencing maltreatment,⁵⁷ thereby perpetuating the intergenerational transmission of trauma. It is also possible that mental health symptoms present at the time of the interview could affect participants’ trauma recollection. Additionally, participants’ recollection of childhood physical and sexual abuse may not be accurate given the time that has elapsed since the events occurred. However, there is evidence that individuals are more likely to underestimate rather than overestimate when recalling childhood abuse,⁵⁸⁹ a phenomenon that would lead to an underestimation of effect sizes. Third, this study did not collect data on certain other key factors that have been identified as potential mediators of the association between maternal history of abuse and child behavior in the literature, including quality and frequency of mother-child interaction, harsh discipline, child maltreatment, and concurrent IPV. Fourthly, potential selection bias is a concern in the population that provided follow-up data and may limit the generalizability of our findings to the broader population of mothers in Lima. Fifth, some subgroup analyses, particularly those for sustained mental health, were based on relatively small sample sizes, which resulted in wide confidence intervals and limited statistical power. Sixth, we acknowledge that evaluating multiple outcomes can increase the potential for Type I error (i.e., spurious significant findings by chance) and that this should be considered when interpreting the results. Finally, while our mediation models are doubly robust and temporality was ensured by the timing of data collection (prenatal exposures were assessed prior to child outcomes) there is likely unmeasured confounding. As a result, our mediation effects should be interpreted as suggestive of potential causal mechanisms rather than definitive proof of causality.

Our findings have several implications for future interventions and implementation research as well as policy and practice. First, prenatal visits may be an important opportunity to identify women who have been exposed to IPV and/or abuse during their childhood and to screen for current depression, anxiety, and PTSD. Such interventions might include not only direct mental health treatment but also parenting support. There is evidence that behavioral coaching improves responsiveness for mothers with depression, but larger studies are needed. ⁵⁹ A number of studies have also demonstrated that perinatal interventions focused on supporting the parent-child relationship can improve caregiving sensitivity.^60,61,62 For women experiencing current IPV, a comprehensive approach is needed that addresses not only the psychosocial consequences of IPV but also IPV itself. The strong association between maternal IPV during pregnancy and PTSD four years later suggests that the situation may persist over time. Emerging evidence has shown that the health system can play a key role in IPV interventions.^63,64,65 Intervention studies are needed to examine the effect of a comprehensive screening and treatment program for trauma-exposed pregnant women on child behavioral outcomes and to incorporate effective preventive and treatment interventions into practice, especially among populations with a high prevalence of child abuse and IPV. Peru’s Ministry of Health has identified perinatal depression as a priority area for service expansion.⁶⁶ The Thinking Healthy Programme, an evidence-based approach to address depression through cognitive behavioral techniques delivered by community health workers without previous mental health training⁶⁷ was launched in Lima by the non-profit organization Socios En Salud. ⁶⁸ Our findings suggest that increased access to such community-based mental health programs would be beneficial, particularly when integrated with violence prevention and parenting services.

This study contributes to the limited literature on the relationship between maternal trauma history, mental health, and child behavior and provides evidence from a low-income population in a middle-income country with a high prevalence of violence against women and children. These results underscore the importance of identifying pregnant women with a history of both IPV and childhood abuse, screening affected women for both depression and PTSD and ensuring that they receive appropriate treatment. Such screening could also help in identifying women who may need greater support with parenting after they give birth to prevent behavioral problems among their children. Our findings suggest that there are multiple opportunities in this cascade to intervene and break the cycle of trauma.