Adverse Childhood Experiences and Socio-emotional Outcomes of Very Preterm Children

Abstract

Objective(s).

To examine whether Adverse Childhood Experiences (ACEs) confers risk for socio-emotional problems in very preterm children (VPT).

Study design.

As part of a longitudinal study, 96 VPT infants (23–30 weeks gestation) were recruited from a level-III neonatal intensive care unit and underwent follow-up at ages 2 and 5 years. Eighty-three full-term (FT, 37–41 weeks gestation) children were recruited from an adjoining obstetric service and the local community. ACEs were assessed with the Child Life Events Scale at age 2 and Preschool Age Psychiatric Assessment at age 5. At age 5, internalizing, externalizing, and Attention-Deficit/Hyperactivity Disorder (ADHD) symptoms were assessed with the Child Behavior Checklist and Conner’s Rating Scale-Revised, respectively. Covariates including socioeconomic disadvantage, maternal distress, and parent ADHD symptoms were assessed at the 2 and/or 5-year follow-up. Mediation and moderation analysis, accounting for family clustering, examined associations between birth group, ACEs, and socio-emotional outcomes.

Results.

After covariate adjustment, VPT children experienced more ACEs (p<0.001), particularly Medical ACEs (p<0.01), and had worse ADHD and internalizing outcomes (p<.05) than FT children. ACEs mediated the association between birth group and ADHD outcomes (95% CIs: 0.11 – 4.08). There was no evidence of mediation for internalizing outcomes. Rather, higher parent ADHD symptoms (p<.001) and maternal distress (p<.05) were associated with poorer internalizing outcomes.

Conclusion(s).

Screening for childhood ACEs should be embedded in the follow-up care of VPT children and their families. Strategies to screen for and address parent psychosocial functioning may be important to support socio-emotional development for both VPT and FT children.

INTRODUCTION

Preterm birth (<37 weeks gestational age [GA]) is a major public-health concern in the United States, with one in ten infants born preterm.^1,2 Despite improvements in antenatal monitoring and neonatal intensive care unit (NICU) practices that have resulted in increased survival rates among infants born very preterm (VPT, <32 weeks GA), VPT children remain at increased risk for longer-term socio-emotional disorders such as Attention-Deficit/Hyperactivity Disorder (ADHD) and anxiety compared to full-term (FT) children.^3,4 One socioenvironmental mechanism that may confer risk for poor socio-emotional outcomes,⁵ but which remains understudied in this population, concerns exposure to Adverse Childhood Experiences (ACEs).⁶

VPT children are more likely to experience ACEs such as parental neglect and emotional, physical, and sexual abuse than FT children.⁷ VPT children also face higher rates of healthcare utilization and hospitalization due to health complications after NICU discharge^8–10 and work in other clinical populations suggests that stressful and traumatic medical experiences may lead to psychological problems in childhood.^11,12 However, little is known regarding the extent that ACEs, either non-medical or medical, relate to socio-emotional problems in VPT children. One study showed that exposure to multiple ACEs placed low-birthweight children at increased need of mental health services,¹³ but the nature and timing of ACEs for different aspects of socio-emotional functioning remains unclear. Such findings may inform when follow-up services should screen for ACEs and provide psychosocial therapies to address stressful and traumatic experiences in VPT children. The aim of this study was to assess the mediating and moderating links between preterm birth and ACEs on socio-emotional outcomes. We hypothesized that ACEs would be more frequent among VPT children, and that ACEs would mediate and/or moderate the link between VPT birth and adverse socio-emotional outcomes at age 5 years.

METHODS

This prospective, longitudinal observational study included VPT (23 – 30 weeks GA) and FT (37 – 41 weeks GA) children (Table 1, Figure 1). See Supplementary Material (pages 3 – 11) for expanded Methods. VPT infants (n=110 at NICU discharge, n=96 retained to 2 – 5-year follow-up) were recruited from the St Louis Children’s Hospital level-III NICU (2007 – 2013). FT control children (n=83) were recruited via two mechanisms. Fifty-four FT children were identified as infants from a contemporaneous study at an adjoining obstetric service (2010 – 2014).¹⁴ In addition, 29 FT control children were recruited from the local community at age 5 years (2014 – 2016).¹⁵ Exclusion criteria spanned chromosomal or congenital abnormality, congenital infection, and parent unable to give informed consent. Exclusion criteria for the FT group also included maternal positive urine drug screen and neonatal acidosis. The Washington University Institutional Review Board approved all procedures. Written informed consent was obtained from all caregivers. At age-corrected 2 years and age 5 years, VPT and FT children completed developmental assessments. Parents completed questionnaires and an interview to obtain information on the child’s life experiences and socio-emotional outcomes.

Table 1.

Infant clinical characteristics and social background at the 5-year follow-up

	Very Preterm	Full-term	X2/t	p
Number of participants at age 2 years	89	32
Number of participants at age 5 years	87	74
Number of participants total from 2 to 5 years	96	83
Infant characteristics
Gestational age, m (SD)	26.59 (1.8)	39.23 (1.1)	58.34	<0.001
Birthweight, m (SD)	943 (257.6)	3,333 (476.0)	40.73	<0.001
Intra-uterine growth restriction, % (n)	2.9 (5)	-
Oxygen at 36 weeks, % (n)	29.1 (52)	-
Patent ductus arteriosus, % (n)	19.0 (52)	-
Necrotizing enterocolitis, % (n)	4.5 (8)	-
Retinopathy of prematurity, % (n)	7.8 (14)	-
Confirmed sepsis, % (n)	17.3 (31)	-
Received postnatal dexamethasone, % (n)	5.0 (9)	-
Did not receive antenatal steroids, % (n)	4.5 (8)	-
Upper quartile for TPN, % (n)	9.7 (17)	-
SDS change greater than −3SD from birth to discharge, % (n)	1.1 (2)	-
High grade white matter injury, % (n)a	15.1 (13)	-
Sex assigned at birth, % (n)
Male	43.8 (42)	43.4 (36)
Female	56.3 (54)	56.6 (47)	0.003	0.96
Child Race and Ethnicity (chart review), % (n)
Hispanic or Latino White	4 (4.2)	2 (2.4)
Hispanic or Latino Black or African American	0 (0.0)	1 (1.2)
Non-Hispanic or Latino Asian	4 (4.2)	0 (0.0)
Non-Hispanic or Latino Biracial	3 (3.2)	1 (1.2)
Non-Hispanic or Latino Black or African American	39 (41.1)	45 (58.5)
Non-Hispanic or Latino White	45 (47.4)	30 (36.6)	9.69	0.08
Ethnicity unknown White	1 (1.2)	0 (1.2)	-	-
Ethnicity unknown Black or African American	0	1	-	-
Maternal Background at the 5-Year Follow-up
Maternal Race and Ethnicity (self-reported), % (n)
Hispanic or Latino White	4 (4.2)	0 (0.0)
Hispanic or Latino Other Race (not defined)	0 (0.2)	2 (2.4)
Non-Hispanic or Latino Asian	4 (4.2)	0 (0.0)
Non-Hispanic or Latino Black or African American	36 (37.5)	49 (59.8)
Non-Hispanic or Latino White	47 (49.0)	31 (37.8)	19.29	.002
Ethnicity unknown White	1 (1.2)	0	-	-
Race and ethnicity unknown	0	1 (1.2)	-	-
Socioeconomic disadvantage component score, m (SD)b	0.07 (0.9)	−0.09 (1.1)	−0.99	0.33
Income to Needs Ratio, m (SD)	1.93 (2.0)	2.29 (2.1)	1.09	0.27
Area Deprivation Index percentile, m (SD)	65.11 (24.6)	59.60 (30.8)	−1.19	0.24
Sociodemographic Stressor Index, m (SD)	1.51 (1.3)	1.59 (1.4)	0.34	0.74
Age at delivery (≤18 years)	4.6 (4)	4.1 (3)	0.03	0.87
Black or African American [proxy for racial discrimination against Black or African American persons in the St. Louis area], % (n) c	37.9 (33)	58.1 (43)	6.53	0.01
Single parent household, % (n)	4.6 (4)	10.8 (8)	2.24	0.14
No High School Qualification, % (n)	47.1 (41)	48.6 (36)	0.04	0.85
Public Health Insurance, % (n)	57.5 (50)	54.1 (40)	0.19	0.66
Psychosocial Distress Index, m (SD)d	0.95 (1.2)	0.97 (1.2)	0.10	0.92
Depression, % (n)	13.1 (11)	9.4 (6)	0.49	0.48
Anxiety, % (n)	16.3 (13)	20.6 (14)	0.46	0.50
Parenting stress, % (n)	15.8 (12)	13.0 (9)	0.22	0.64
Low social support, % (n)	28.6 (22)	34.3 (24)	0.56	0.46
Adverse family life events, % (n)	23.7 (18)	17.1 (12)	0.96	0.33
Maternal ADHD symptom t-score, m (SD)	42.76 (9.8)	41.88 (11.3)	−0.49	0.62
Paternal ADHD symptom t-score, m (SD)	48.02 (11.3)	45.38 (12.4)	−1.14	0.26

Note. TPN, Total parenteral nutrition; SDS, Standard deviation score

^aDefined as grade III/IV intraventricular hemorrhage, post-hemorrhagic hydrocephalus, periventricular leukomalacia on term-equivalent age anatomical MRI scan

^bComposite of demographic stressor index, Income to Needs Ratio (z-scored and reversed scored), and area deprivation index

^cVariable coding: Black or African American = 1, Non-Black = 0

^dSee Supplemental Material for complete details

Figure 1.

Participant flow from recruitment to follow-up in very preterm (VPT) and full-term (FT) children.

Measures

ACEs.

ACEs were assessed at age 2 years using the parent-report Child Life Events Scale (CLES),¹⁶ obtained for 117 (97%) of children. At age 5 years, ACEs were assessed using the Life Events List A (past 6 months) and List B (lifetime) from the parent-report Preschool Age Psychiatric Assessment (PAPA),¹⁷ a semi-structured diagnostic interview obtained for 123 (76%) of children. During the PAPA interview, ages/dates, and written information were used to determine when the ACE occurred (i.e., ages 0 – 2 or 2 – 5 years). For children who completed both the 2- and 5-year follow-up, given dates/ages were used to ensure there was no overlap in ACEs. See Supplemental Material (page 13) for ACEs included in this study. In addition to total ACEs, ACEs timing variables were created at ages 2 and 5 years. At age 2 years, this variable consisted of ACEs between birth and age 2 years on the CLES and PAPA List B if the event occurred from age 0 – 2 years. For children who only completed the 5-year visit, PAPA List B captures similar ACEs as the CLES. At age 5 years, this variable consisted of ACEs between age 2 and 5 years on the PAPA List A and List B if the event occurred from age 2 – 5 years.

ACEs Subtypes.

Potentially Traumatic Events (PTEs) were defined as a subset of serious and traumatic ACEs likely to cause intense fear or harm.¹⁸ Medical ACEs included medical procedure, operation, emergency department visit, or overnight hospitalization. The remaining ACEs were categorized as Non-Medical ACEs. Timing variables from birth to age 2 years and from age 2 to 5 years were also created for PTEs, Medical ACEs, and Non-Medical ACEs.

Socio-emotional Outcomes.

At the 5-year follow-up, parents completed the Child Behavior Checklist (CBCL)¹⁹ and Conner’s Rating Scale-Revised (CRS-R)²⁰ to assess children’s internalizing and externalizing symptoms (n=153 obtained, 95%) and ADHD symptoms (n=153 obtained, 95%), respectively. Higher t-scores indicate greater symptomology.

Infant Factors.

Gestational age, birthweight, sex assigned at birth (male=1, female=2), race and ethnicity, and neonatal morbidities were obtained from medical records at birth. The neonatal illness index²¹ was based upon the presence of intrauterine growth restriction, prolonged oxygen supplementation, did not receive antenatal steroids, received dexamethasone, necrotizing enterocolitis, sepsis, patent ductus arteriosus, retinopathy of prematurity, ≥3 SD decrease in weight-for-height/length to term-equivalent age, and >75th percentile for parenteral nutrition.

Psychosocial Factors.

Maternal race and ethnicity were self-reported. Social and family factors were obtained from the 2- and 5-year follow-up (see Supplemental Material for expanded details). Briefly, for each timepoint, Principal Component Analysis was used to create a multidimensional²² socioeconomic disadvantage score²³ including factors such as maternal sociodemographic stressors (Table 1),^15,24 neighborhood Area Deprivation Index,²⁵ and family Income to Needs Ratio.²⁶ The 5-year maternal Psychosocial Distress Index^21,24 included the Beck Depression Inventory-II,²⁷ State Trait Anxiety Inventory,²⁸ Parent Stress Index (PSI),²⁹ Social Support Questionnaire,³⁰ and PSI Life Events faced by any family member. Mothers completed the self and observer CRS-R to obtain maternal and paternal ADHD symptom t-scores.

Data Analysis

Analysis was performed in SPSS (Version 29, IBM). After screening for outliers (>3 SD, Table S1), linear-mixed effect models (LMMs) examined birth-group differences (VPT=1, FT=0) in ACEs and socio-emotional outcomes with family clustering included as a random effect with random intercept. LMMs were extended to account for socioeconomic disadvantage, maternal Psychosocial Distress Index, and parent ADHD symptoms (see Table S2). For mediation analyses, direct and indirect effects (95% confidence intervals [CIs] calculated with Monte Carlo sampling) were obtained using MLMED 2.0³¹ for nested data. Moderation analysis with main effects and mean-centered interaction terms was performed using LMMs with bootstrapping. Supplementary mediation/moderation was also performed with ACEs in early childhood.

RESULTS

Stressful and Traumatic Events

ACEs.

VPT children experienced more total ACEs than FT children (p<0.001), especially from birth to age 2 years (p=0.01). Between-groups differences in total ACEs persisted after covariate adjustment for social and family background factors (p=.007, Table 2; see also Table 3 for frequently reported ACEs).

Table 2.

Birth group comparisons in Adverse Childhood Events (ACEs), Medical ACEs, Non-Medical ACEs, and Potentially Traumatic Events (PTEs) from birth to age 5 years

m (SD)	Very Preterm	Full-term	Cohen’s d	t	p a	p b
ACEs
Birth to age 2 yearsc	2.11 (2.2)	1.12 (1.1)	0.57	−2.53	0.01 *	0.17
Age 2 to 5 yearsd	5.23 (3.7)	4.15 (2.9)	0.32	−1.61	0.11	0.30
Totald	7.13 (4.6)	4.40 (3.0)	0.70	−3.42	<0.001 *	0.007 *
Medical ACEs
Birth to age 2 years	1.14 (1.8)	0.24 (0.50)	0.68	−2.77	0.007 *	0.07
Age 2 to 5 years	0.64 (1.5)	0.18 (0.38)	0.45	−2.07	0.04 †	0.02 †
Total	1.71 (2.5)	0.23 (0.4)	0.83	−3.85	<0.001 *	<0.001 *
Non-medical ACEs
Birth to age 2 years	1.00 (1.2)	0.91 (1.2)	0.08	−.0.61	0.54	0.85
Age 2 to 5 years	4.71 (3.1)	4.10 (3.1)	0.20	−0.98	0.33	0.62
Total	5.54 (3.6)	4.25 (3.1)	0.38	−1.87	0.07	0.21
PTEs
Birth to age 2 years	0.19 (0.5)	0.15 (0.5)	0.08	−0.58	0.56	0.88
Age 2 to 5 years	0.29 (0.6)	0.38 (0.6)	0.08	0.79	0.43	0.63
Total	0.42 (0.8)	0.45 (0.7)	0.04	0.11	0.92	0.94

^aValue from linear mixed-effects models with family clustering

^bValue from linear mixed-effects models with family clustering, and covariate factors described in ^c and ^d

^cBirth to age 2 years comparisons adjusted for 2-year socioeconomic disadvantage and maternal Psychosocial Distress Index, and parent ADHD symptoms (results were identical with or without parent ADHD symptoms collected at the 5-year visit)

^dAge 2 to 5 years and Total comparisons adjusted for 5-year socioeconomic disadvantage, maternal Psychosocial Distress Index, and parent ADHD symptoms

^*p-value passed multiple comparison correction using BH-FDR q < .05.

^†p-value passed multiple comparison correction using BH-FDR q < .10

Table 3.

Most frequently endorsed Adverse Childhood Events (ACEs) and Potentially Traumatic Events (PTEs) on the CLES and the PAPA

	VPT		FT		Odds ratio (95% CI)a	p a
	%	n	%	n
Top 5 ACES
CLES
Overnight hospitalization or medical operation(s)	52.9	46	21.2	7	4.17 (1.58 – 10.98)	.004 †
Seen or heard adult family members arguing very loudly or fighting	21.2	18	15.6	5	1.45 (0.48 – 4.41)	0.51
Separated from parent or guardian ≥ 1 week	21.2	18	15.6	5	1.45 (0.46 – 4.54)	0.52
Had a loved one die	15.3	13	21.9	7	0.65 (0.22 – 1.90)	0.43
Gotten a new baby brother or sister	10.6	9	12.5	4	0.83 (0.23 – 3.03)	0.77
PAPA
Change in school/daycare	63.9	53	47.5	19	1.95 (0.88 – 4.35)	0.10
Moved house	55.6	45	57.5	23	0.92 (0.41 – 2.09)	0.85
Hospitalization, ED visit, or operation	28.9	24	15.0	6	2.32 (0.83 – 6.40)	0.11
Death of an adult loved one	27.7	23	32.5	13	0.80 (0.33 – 1.94)	0.62
New child in home	30.1	25	32.5	13	0.90 (0.38 – 2.12)	0.80
Top 5 PTEs
CLES
Been bitten by a dog	7.1	6	6.3	2	1.14 (0.22 – 6.00)	0.88
Seen someone hit, push, or kick a family member	5.9	5	3.1	1	1.94 (0.22 – 17.34)	0.55
Been seriously hurt	3.5	3	0.0	0	1.04 (1.00 – 1.08)	0.56b
Seen someone use a weapon to threaten or hurt a family member	1.2	1	0.0	0	1.01 (0.99 – 1.04)	1.00b
Been in a car accident	1.2	1	3.1	1	0.37 (0.02 – 6.10)	0.49
PAPA
Witness an event that caused harm or death to another	9.8	8	2.5	1	4.22 (0.45 – 39.20)	0.21
Poisoning	4.8	4	0.0	0	1.05 (1.00 – 1.10)	0.30b
Natural disaster (e.g., storm, tornado, earthquake)	3.6	3	7.5	3	0.46 (0.09 – 2.42)	0.36
Accidental burning (> 1st degree burn)	3.6	3	7.5	3	0.46 (0.09 – 2.39)	0.36
Physical abuse	2.4	2	0.0	0	1.03 (0.99 – 1.06)	1.00b

^aBinomial logistic regression using generalized estimated equations with family clustering for dichotomous dependent variables

^bFisher’s exact test due to 0% in FT group, generalized estimated equations did not converge

^†p-value passed multiple comparison correction using BH-FDR q < .10

PTEs.

There were no between-groups differences in PTEs at any timepoint (all p>.05, Table 2), potentially due to the low frequency of PTEs (Table 3). Because of the overall low frequency and lack of between-groups differences in PTEs, PTEs were not examined in subsequent mediation/moderation analyses.

Medical ACEs.

VPT children experienced more Medical ACEs across childhood than FT children (all p<.05). Findings for total Medical ACEs, as well as Medical ACEs from 2 to 5 years, persisted after covariate adjustment (p<0.05, Table 2). Supplementary bi-variate correlations between infant clinical factors and subsequent Medical ACEs within the VPT group are shown in Supplementary Material (page 19).

Non-Medical ACEs.

There were no between-groups differences in Non-Medical ACEs at any timepoint (all p>.05, Table 2). Thus, Non-Medical ACES were not examined in subsequent mediation/moderation analyses.

Socio-emotional Outcomes at Age 5 Years

VPT children had higher levels of ADHD symptoms (p=0.02), internalizing symptoms (p=0.03), and overall socio-emotional problems (p=0.04), with between-groups differences persisting after covariate adjustment (p<.05, Table 4). In contrast, there was no between-groups difference in externalizing problems (p=.13).

Table 4.

Mental health outcomes of very preterm and full-term children at age 5 years

m (SD)	Very Preterm	Full-term	Cohen’s d	t a	p a	p b
CRS-R ADHD index	57.52 (14.7)	52.04 (12.8)	0.40	−2.31	0.02 *	0.04 †
CBCL internalizing	49.35 (14.3)	44.61 (10.7)	0.38	−2.26	0.03 *	0.02 *
CBCL externalizing	47.83 (14.4)	44.23 (13.7)	0.26	−1.51	0.13	0.08
CBCL overall socio-emotional problems	48.83 (15.5)	43.79 (13.2)	0.35	−2.08	0.04 †	0.02 *

Note. CRS-R, Conners’ Rating Scales-Revised; ADHD, Attention Deficit Hyperactivity Disorder; CBCL, Child Behavior Checklist. T-scores used for all outcomes.

^aValue from linear mixed-effects models with family clustering

^bValue from linear mixed-effects models with family clustering, and 5-year covariate factors socioeconomic disadvantage, maternal Psychosocial Distress Index, and parent ADHD symptoms

^*p-value passed multiple comparison correction using BH-FDR q < .05.

^†p-value passed multiple comparison correction using BH-FDR q < .10

Mediation Analyses

Total ACEs.

As shown in Table 5, total ACEs mediated the association between VPT birth and ADHD symptoms at age 5 years (95% CIs: 0.57 – 5.40). Mediation results for ADHD symptoms remained significant (Figure 2) when socioeconomic disadvantage (B=0.66, p=0.60), maternal Psychosocial Distress Index (B=5.16, p<0.001), and parent ADHD symptoms (B=4.14, p=0.03) were included as covariate factors.

Table 5.

Mediation analysis of birth group, total Adverse Childhood Events (ACEs) from birth to age 5 years, and mental health outcomes at age 5 years

		Unadjusted estimates			Estimates adjusted for covariate factorsa
Path	Description	B (SE)	p	95% CIs	B (SE)	p	95% CIs
Birth group, total ACEs, ADHD symptoms
Path A	Birth group → total ACEs	2.79 (0.82)	0.001	1.17 – 4.41	2.29 (0.83)	0.007	0.65 – 3.93 b
Path B	Total ACEs → ADHD symptoms controlling for birth group	0.94 (0.34)	0.007	0.27 – 1.61	0.73 (0.32)	0.03	0.08 – 1.37 b
Path C	Birth group → ADHD symptoms	6.70 (2.73)	0.02	1.28 – 12.12	5.20 (2.44)	0.04	0.36 – 10.04 b
Path C’	Birth group → ADHD symptoms controlling for total ACEs	4.17 (2.81)	0.14	−1.40 – 9.74	3.37 (2.56)	0.19	−1.74 – 8.47 b
ab	Birth group → ADHD symptoms through total ACEs	2.62 (1.25)	-	0.57 – 5.40	1.67 (0.99)	-	0.11 – 4.08 b
Birth group, total ACEs, Internalizing problems
Path A	Birth group → total ACEs	2.79 (0.82)	0.001	1.17 – 4.41	2.29 (0.83)	0.007	0.65 – 3.93
Path B	Total ACEs → Internalizing controlling for birth group	0.51 (0.32)	0.12	−0.14 – 1.15	0.31 (0.26)	0.24	−0.21 – 0.84
Path C	Birth group → Internalizing	6.21 (2.59)	0.02	1.08 – 11.33	4.34 (2.00)	0.03	0.38 – 8.31
Path C’	Birth group → Internalizing controlling for total ACEs	4.94 (2.71)	0.07	−0.43 – 10.31	3.71 (2.10)	0.08	−0.47 – 7.90
ab	Birth group → Internalizing through total ACEs	1.42 (1.03)	-	−0.34 – 3.69	0.72 (0.70)	-	−0.49 – 2.39
Birth group, total ACEs, Overall Socio-emotional Problems
Path A	Birth group → total ACEs	2.79 (0.82)	0.001	1.17 – 4.41	2.29 (0.83)	0.007	0.65 – 3.93
Path B	Total ACEs → Socioemotional controlling for birth group	0.49 (0.36)	0.17	−0.22 – 1.21	0.28 (0.29)	0.35	−0.31 – 0.86
Path C	Birth group → Socio-emotional	7.29 (2.86)	0.01	1.63 – 12.96	5.68 (2.19)	0.01	1.32 – 10.03
Path C’	Birth group → Socio-emotional controlling for total ACEs	6.00 (3.00)	0.04	0.05 – 11.95	5.14 (2.32)	0.03	0.52 – 9.74
ab	Birth group → Socio-emotional through total ACEs	1.38 (1.12)	-	−0.55 – 3.83	0.63 (0.75)	-	−0.72 – 2.42
Birth group, medical ACEs, ADHD Symptoms
Path A	Birth group → medical ACEs	1.52 (0.39)	<0.001	0.74 – 2.30	1.40 (0.33)	<0.001	0.73 – 2.06
Path B	Medical ACEs → ADHD symptoms controlling for birth group	1.04 (0.79)	0.19	−0.53 – 2.60	0.79 (0.85)	0.36	−0.90 – 2.49
Path C	Birth group → ADHD symptoms	6.70 (2.73)	0.02	1.28 – 12.12	5.20 (2.44)	0.04	0.36 – 10.04
Path C’	Birth group → ADHD symptoms controlling for medical ACEs	5.00 (2.99)	0.10	−0.94 – 10.95	3.98 (2.80)	0.16	−1.60 – 9.56
ab	Birth group → ADHD symptoms through medical ACEs	1.57 (1.30)	-	−0.73 – 4.34	1.10 (1.25)	-	−1.28 – 3.83
Birth group, medical ACEs, Internalizing Problems
Path A	Birth group → medical ACEs	1.52 (0.39)	<0.001	0.74 – 2.30	1.40 (0.33)	<0.001	0.73 – 2.06
Path B	Medical ACEs → Internalizing controlling for birth group	1.19 (0.84)	0.16	−0.47 – 2.85	0.52 (0.68)	0.45	−0.83 – 1.88
Path C	Birth group → Internalizing	6.21 (2.59)	0.02	1.08 – 11.33	4.34 (2.00)	0.03	0.38 – 8.31
Path C’	Birth group → Internalizing controlling for medical ACEs	4.69 (2.81)	0.10	−0.88 – 10.26	3.71 (2.24)	0.10	−0.75 – 8.17
ab	Birth group → Internalizing through medical ACEs	1.80 (1.39)	-	−0.62 – 4.78	0.73 (0.99)	-	−1.19 – 2.86
Birth group, medical ACEs, Overall Socio-emotional Problems
Path A	Birth group → medical ACEs	1.52 (0.39)	<0.001	0.74 – 2.30	1.40 (0.33)	<0.001	0.73 – 2.06
Path B	Medical ACEs → Socio-emotional controlling for birth group	1.29 (0.93)	0.17	−0.56 – 3.13	0.69 (0.76)	0.37	−0.83 – 2.20
Path C	Birth group → Socio-emotional	7.29 (2.86)	0.01	1.63 – 12.96	5.67 (2.19)	0.01	1.32 – 10.03
Path C’	Birth group → socio-emotional controlling for medical ACEs	5.59 (3.10)	0.07	−0.57 – 11.75	4.82 (2.46)	0.05	−0.08 – 9.72
ab	Birth group → socio-emotional through medical ACEs	1.95 (1.54)	-	−0.74 – 5.23	0.96 (1.11)	-	−1.16 – 3.38

Note. PTEs, Non-Medical ACEs, and externalizing problem outcomes were not examined due to no significant birth-group differences on these variables. Path C, total effect; Path C’, direct effect; ab, indirect effect. Significance of indirect effect determined from Monte Carlo 95% confidence intervals (10,000 samples). Unstandardized estimates (B) shown. Family clustering is included in the estimation of direct and indirect effects using MLMED. As MLMED does not generate total effects, total effects were estimated using PROCESS.

^aIncludes 5-year socioeconomic disadvantage (p>.05 in all models), maternal Psychosocial Distress Index (p<.05 in all models), and parent ADHD symptoms (p<.05 in all models) concurrent with mental health outcomes.

^bModel estimates shown in Figure 2.

Figure 2.

Summary of mediation analysis showing that the association between birth group and ADHD outcomes at age 5 years (Path C) was explained (Path C’) by total ACEs from birth to age 5 years (indirect effect 95% confidence intervals [CI] 0.11 – 4.08). Unstandardized estimates (B) shown. Model accounts for family clustering and covariate adjustment for 5-year socioeconomic disadvantage, maternal Psychosocial Distress Index, and parent ADHD symptoms.

Total ACEs did not mediate the association between VPT birth and internalizing or overall socio-emotional problems (all 95% CIs crossed zero, full null results available upon request). Rather, maternal Psychosocial Distress Index (B range 2.31 – 2.75, p<.05) and parent ADHD symptoms (B range 8.52 – 9.64, p<.001) were found to be the only covariate factors independently related to internalizing and overall socio-emotional problem outcomes over and above ACEs (p>.05) in multivariate mediation models.

Supplementary mediation analysis.

Given that VPT children experienced more ACES relative to FT children from birth to age 2 years (d = 0.57) than from age 2 to 5 years (d = 0.32, Table 2), a second supplementary mediation analysis was performed using ACEs from birth to age 2 years as the mediator, showing that indirect effects were significant for internalizing (95% CIs: 0.22 – 4.54) and overall socio-emotional problems (95% CIs: 0.19 – 4.95) but results did not persist after covariate adjustment (Table S3).

Medical ACEs.

As shown in Table 5, total Medical ACEs did not explain or mediate the birth group differences mental health outcomes at age 5 years, either before or after covariate adjustment for social and family factors.

Moderation Analysis

There were no interactions between birth group and either total ACEs or Medical ACES from birth to age 5 years on ADHD symptoms, internalizing, or overall socio-emotional problems. There were also no interactions between birth group and maternal Psychosocial Distress Index or socioeconomic disadvantage, or between maternal Psychosocial Distress Index and socioeconomic disadvantage, on mental health outcomes (all interaction terms p>.05, null results available in full upon request).

Supplementary moderation analysis.

A second supplementary moderation analysis was performed using ACEs from birth to age 2 years as the moderator. As shown in Table 6 including the main effects of birth group (p=.03) and ACEs from birth to age 2 years (p=.69), there was an interaction between birth group and ACEs from birth to age 2 years on ADHD symptoms at age 5 years (p=.049). Supplementary within-group correlations demonstrated that the association between ACEs and ADHD symptoms was significant in the VPT group (r=.28, p=.02) but not in the FT group (r=.13, p=.54). However, the interaction term was not significant after covariate adjustment (p=.13, Table 6).

Table 6.

Summary of moderation analysis suggesting that the interaction between birth group and early childhood ACEs on ADHD symptom outcomes is attenuated after accounting for maternal distress and parent ADHD symptoms.

	Model 1			Model 2
	Pseudo R2 = .10			Pseudo R2 = .54
	B (SE)	95% CIs	p	B (SE)	95% CIs	p
Birth group	6.91 (3.34)	1.55 – 11.78	0.03	8.05 (2.99)	2.63 – 12.18	0.005
ACEs birth to age 2 years	0.31 (0.98)	−1.08 – 1.81	0.69	−0.70 (1.04)	−2.06 – 1.30	0.45
Interaction: Birth group x ACEs birth to age 2 years	3.29 (2.09)	0.07 – 6.39	0.049	2.90 (2.15)	−1.19 – 5.77	0.13
Socioeconomic Disadvantage	-		-	0.19 (1.33)	−2.04 – 2.38	0.89
Maternal Psychosocial Distress Index	-		-	5.28 (1.49)	2.91 – 7.93	<0.001
Parent ADHD Symptoms	-		-	4.38 (2.04)	1.49 – 8.10	0.03

Note. Estimates, confidence intervals, and p-values from linear mixed-effects models performed with percentile-based bootstrapping (1000 samples) and accounting for family clustering. Mean-centered interaction terms used. Covariate social and family factors from 5-year follow-up concurrent with mental health outcomes.

DISCUSSION

After covariate adjustment for social and family background factors, VPT children experienced more ACEs compared to FT children by age 5 years. Medical ACEs were also found to be more common in VPT children, especially for those with lower GA and/or birthweight. These findings are consistent with studies showing that VPT children often have ongoing medical needs after NICU discharge along with poorer health outcomes that may require increased healthcare utilization or hospitalization.^10,32 Taken together, VPT infants born at lower GA and/or birthweight and their caregivers may need additional surveillance post NICU-discharge to reduce the likelihood of emergency department or urgent care visits as well as extra support when frequently navigating the healthcare system. In contrast, rates of PTEs and Non-Medical ACEs were similar between VPT and FT children. Other socioenvironmental factors including socioeconomic disadvantage and maternal distress were instead found to be correlated with exposure to Non-Medical ACEs and PTEs. This finding may reflect the presence of risk factors such as environmental instability or unpredictability, family conflict or other related stressors, and exposure to unsafe neighborhoods in lower socioeconomic settings, and in turn, increased likelihood of exposure to ACEs.^33–36

At age 5 years, VPT children had higher levels of ADHD symptoms, internalizing problems, and overall socio-emotional problems than FT children, with these differences persisting after covariate adjustment. These findings are consistent with the Preterm Behavioral Phenotype^4,37,38 and prior work in this cohort.^3,21,24 Building upon our previous work as well as existing literature reporting associations between ACEs and ADHD in children and adolescents,³⁹ we found that total ACEs throughout childhood served as a key mediating factor linking VPT birth with increased ADHD symptoms, even after accounting for socioeconomic disadvantage, maternal distress, and parent ADHD symptoms. This suggests that cumulative exposure to different types of ACEs may increase risks of ADHD in VPT children, independent of other salient environmental and heritable risk factors readily implicated in child psychopathology. These findings highlight that exposure to ACEs in the first five years of life, when the immature brain is highly sensitive to environmental exposures, may be particularly impactful for adverse mental health outcomes.^40–42 Increased stress during periods of rapid brain development can lead to alterations in the stress-regulatory neural circuits involving the hippocampus and amygdala, which are key brain regions frequently implicated in the development of psychopathology.^43–45 The glucocorticoid-rich amygdala and hippocampus undergo significant development in the first years of life, making them especially sensitive to early adversity.^46–48 Adversity-related alterations in these circuits also have secondary consequences for disrupted Hypothalamic-Pituitary-Adrenal axis function and cortisol production, together contributing to the underlying neurobiology of internalizing and externalizing problems.^45,49,50 Thus, VPT children may be particularly susceptible to ACEs as a mechanism towards later ADHD problems, given that VPT infants show altered amygdala and hippocampal development^51–54 and are thought to be more sensitive or susceptible to environmental exposures.^55,56

Although VPT birth was associated with a greater incidence of Medical ACEs, there was no evidence to suggest that Medical ACEs explained the poorer mental health outcomes among VPT children. This finding was unexpected, given that medical procedures may be frightening and painful for children, and hospitalization often results in separation from primary caregivers as well as disrupted routines, resulting in increased risk for child depression and anxiety.^57–60 These previous studies have focused on hospitalized populations,^57,60 whereas we also included out-patient services such as emergency department visits in our Medical ACEs variable. Medical ACEs may also be indirectly related to mental health outcomes, with other research suggesting that being born preterm is associated with prenatal exposure to maternal inflammation as well as sustained postnatal activation of proinflammatory responses that, together, may impair the immune system leading to poor health outcomes.^61–63 In turn, chronic activation of proinflammatory responses has been broadly implicated in psychiatric disorders including depression,^64,65 potentially through the effects of proinflammatory cytokines on the developing brain.^66–69

Maternal distress and parent ADHD symptoms were important covariate factors associated with worse internalizing and overall socio-emotional problems at age 5 years. Parent ADHD symptoms and maternal mood/affective problems have been linked to internalizing problems in preschool and school-age children in numerous studies.^70,71 Relationships between parent psychosocial dysfunction and internalizing problems in VPT children could be transmitted directly via shared heritable liabilities in socio-emotional dysregulation common to multiple forms of psychopathology^72,73 or indirectly via other postnatal exposures associated with both parent ADHD symptoms and child psychopathology such as exposure to harsh and intrusive parenting behaviors.^74,75 The extent to which associations between ACEs and exposure to related psychosocial stressors on mental health outcomes might vary as a function of other moderating factors such as socioeconomic disadvantage remains unclear, as larger samples would be needed to understand more these more complex mechanistic pathways towards psychopathology.

Strengths of this study include the prospective, longitudinal study design with a high retention rate in the VPT group, the inclusion of demographically similar FT children, repeat assessment of ACEs, and the consideration of key covariate factors known to be related to VPT birth, ACEs, and psychopathology. Study limitations include reliance on primary caregiver report of child socio-emotional problems. The use of different measures to assess ACEs and the inclusion of two largely non-overlapping FT control groups at the 2- and 5-year follow-up assessments precluded the ability to conduct intra-individual longitudinal analyses and thus our results are cross-sectional. However, we note that the CLES and PAPA were age-appropriate measures and, overall, capture similar types of stressful and traumatic experiences. We also acknowledge that caregivers may have under-reported serious adverse events concerning child neglect or abuse, due to fear of mandatory reporting to Child Protection Services. As such, results may not generalize to VPT children who have experienced major adverse events. We did not have a direct measure of racial discrimination available to include in our Sociodemographic Stressor Index. Additionally, given that our cohort was recruited from a single-site tertiary care center in the Midwestern United States, and rates of preterm birth reflect both the medical care and socio-cultural environment of the region of recruitment, our findings may not generalize to other samples of VPT children. Sample size of the FT group at the 2-year follow-up was modest and we may have been under-powered to detect small-medium between-groups differences in 2-year ACEs. To address these limitations, future research should include secondary caregiver reports of children’s life experiences and socio-emotional outcomes and an objective assessment of severe PTEs from child welfare records.

Findings of the current study have important research and clinical implications. Future research examining ADHD outcomes of VPT children and adolescents should account for stressful and traumatic life experiences, especially during early childhood, as part of the developmental pathway towards psychopathology. Given that VPT children are more likely to experience Medical ACEs than FT children, embedding strategies to reduce stress and trauma during medical care and hospitalization may be important for VPT children. Existing intervention-based studies have shown that specialized staff training, psycho-educational preparation, therapeutic play, mother-child interaction therapy, and peer support programs are related to better socio-emotional and medical outcomes in hospitalized children.^76–79 Providing education and support programs for families after hospital-discharge may also support VPT children’s longer-term socio-emotional outcomes.^80,81

CONCLUSIONS

VPT children experienced higher rates of ACEs, particularly during early childhood, compared to their FT peers. The increased prevalence of ACEs was found to mediate the association between VPT birth and increased ADHD symptoms by age 5 years. As such, screening for ACEs should be embedded in the routine, longer-term follow-up care of VPT infants and children. Additionally, psychiatric clinical care treating ADHD in VPT children should also include psychotherapies to address the stress and trauma associated with ACEs and as well as exposure to psychosocial adversity. Parent ADHD symptoms and maternal psychosocial distress were also found to be key aspects of the socio-familial environment related to child internalizing and overall socio-emotional outcomes. Strategies to screen for and address parent psychosocial functioning may also be important to include as part of routine health care to support socio-emotional development for all children regardless of GA at birth.

LIST OF ABBREVIATIONS AND ACRONYMS

ACEs

Adverse Childhood Experiences

ADHD

Attention-Deficit/Hyperactivity Disorder

ADI

Area Disadvantage Index

CBCL

Child Behavior Checklist

CI

Confidence Interval

CLES

Child Life Events Scale

CRS-R

Conner’s Rating Scale-Revised

FT

Full-term

GA

Gestational Age

LMMs

Linear mixed-effects models

NICU

Neonatal intensive care unit

PAPA

Preschool Age Psychiatric Assessment

PTEs

Potentially traumatic events

VPT

Very preterm

Data Sharing Statement:

Data can be made available to qualified researchers by written request to Drs. C. D. Smyser and C. E. Rogers under the guidance of a formal data sharing agreement. Data related to the larger study will be shared on the open access NIMH Data Archive repository and made publicly available at the conclusion of this study.