Systematic review of pediatric health outcomes associated with childhood adversity

Background Early detection of and intervention in childhood adversity has powerful potential to improve the health and well-being of children. A systematic review was conducted to better understand the pediatric health outcomes associated with childhood adversity. Methods PubMed, PsycArticles, and CINAHL were searched for relevant articles. Longitudinal studies examining various adverse childhood experiences and biological health outcomes occurring prior to age 20 were selected. Mental and behavioral health outcomes were excluded, as were physical health outcomes that were a direct result of adversity (i.e. abusive head trauma). Data were extracted and risk of bias was assessed by 2 independent reviewers. Results After identifying 15940 records, 35 studies were included in this review. Selected studies indicated that exposure to childhood adversity was associated with delays in cognitive development, asthma, infection, somatic complaints, and sleep disruption. Studies on household dysfunction reported an effect on weight during early childhood, and studies on maltreatment reported an effect on weight during adolescence. Maternal mental health issues were associated with elevated cortisol levels, and maltreatment was associated with blunted cortisol levels in childhood. Furthermore, exposure to childhood adversity was associated with alterations of immune and inflammatory response and stress-related accelerated telomere erosion. Conclusion Childhood adversity affects brain development and multiple body systems, and the physiologic manifestations can be detectable in childhood. A history of childhood adversity should be considered in the differential diagnosis of developmental delay, asthma, recurrent infections requiring hospitalization, somatic complaints, and sleep disruption. The variability in children’s response to adversity suggests complex underlying mechanisms and poses a challenge in the development of uniform diagnostic guidelines. More large longitudinal studies are needed to better understand how adversity, its timing and severity, and the presence of individual genetic, epigenetic, and protective factors affects children’s health and development.


Background
Evidence has shown that adversities such as abuse or neglect experienced during childhood can have lifelong consequences by affecting the foundations of health [1][2][3][4]. This body of evidence has generated growing attention to the concept of toxic stress, the chronic or frequent activation of the stress response from exposure to serious childhood adversity in the absence of adequate support or protection from adults [5]. Through a mechanism that is influenced by genetic, social, and biological factors, exposure to childhood adversity has been linked to the dysregulation of the neuroendocrine immune circuitry, which results in alterations of brain architecture and other organ systems during sensitive periods of development [6,7]. Negative health outcomes that follow later in life reflect the physiological, epigenetic, and cognitive consequences that the brain and body pay for adapting to stressful and traumatic experiences [2,8].
In one of the earliest studies examining the association between cumulative childhood adversity and health outcomes, known as the Adverse Childhood Experiences (ACE) Study, Felitti et al [9] found a dose-response relationship between ACEs and many leading causes of death in adults, such as chronic respiratory disease, stroke, heart disease, cancer, and diabetes. A growing body of research has since further confirmed that childhood adversity is associated with chronic disease and early death [10][11][12][13][14][15]. Hence, early detection and intervention can have a positive, lifelong impact on an individual's health and well-being. Current efforts to address toxic stress in children are limited by gaps in understanding of the biological response to childhood adversity, and much of the existing evidence is based on adult studies with retrospective reports of adversity in childhood [15][16][17]. Studies on childhood adversity with pediatric health outcomes have pointed to an increased likelihood of physical and developmental health issues in children exposed to adversity [18][19][20], but these studies have been primarily cross-sectional, making it difficult to determine the temporal relationship between exposure and outcome. The present systematic review, therefore, was focused on longitudinal studies in children to better understand the biological mechanisms linking exposure to childhood adversity with pediatric health outcomes.

Methods
We followed the PRISMA Statement [21] in the conduct and reporting of this systematic review.

Data sources
PubMed, PsycArticles, and CINAHL were searched for the following terms in the title or abstract: ("early life" OR adolescent OR child OR infant OR youth OR childhood OR prenatal OR "in utero") AND (divorce OR "parental incarceration" or "parental depression" OR abuse OR neglect OR adversity OR maltreatment OR "toxic stress" OR "allostatic load" OR "adverse childhood experience"). Other terms relevant to childhood adversity such as "trauma," "parental mental health," and "parental substance abuse" were not included as search terms because they were not specific enough to identify additional relevant articles. The searches were limited to full-text articles in the English language, published between January 1, 2001 and December 31, 2015. In addition, the searches excluded animal studies, case reports, review articles, and qualitative studies.

Study selection
Title and abstracts were reviewed and nonrelevant records (studies with no exposure to childhood adversity and/or biological health outcomes) were excluded. Exposure to childhood adversity was defined as exposure to 1 or more of the experiences shown in Table 1, which were derived from the Center for Youth Wellness ACE Questionnaire (CYW ACE-Q) [22]. The CYW ACE-Q was designed to be administered to children aged 0-19 years and includes the traditional 10 ACEs from the ACE Study, as well as additional early life stressors identified by experts and community stakeholders. An adverse exposure in-utero (e.g. maternal drug use) was part of the original search criteria, but these studies were subsequently excluded due to conceptual differences in exposures and outcomes. Biological health outcomes were defined as objective and specific developmental or clinical changes in the health of children. Mental and behavioral health outcomes were excluded, as were direct and immediate physical effects of adverse incidents (i.e. abusive head trauma).
The first two authors (D.O. and P.J.) evaluated the full text of the retained records using eligibility criteria shown in Fig. 1. Subsequently, only records of prospective and retrospective cohort studies examining child outcomes occurring prior to the age of 20 years were retained. The retained records were further evaluated using the following inclusion criteria: 1) robust ascertainment of an ACE exposure, as defined in Table 1; 2) robust ascertainment of a biological health outcome, excluding general measures of health, health-related quality of life, and death; 3) exposure measured prior to outcome; and 4) use of an unexposed comparison group. Each study was reviewed by the 2 authors, and disagreements were resolved through discussion. As a final step in study selection, the references of the selected studies were searched for additional studies that fit all inclusion criteria.

Data extraction and synthesis
We extracted data on setting, study design, sample size, sample description, length of follow-up, adverse exposures, biological health outcomes, and relevant main findings (Tables 2-3). When both simple and adjusted analyses were reported, we extracted the findings that adjusted for covariates and/or confounders. The data were narratively synthesized and organized by type of biological outcome.

Bias assessment
The first two authors (D.O. and P.J.) independently assessed the risk of selection bias, information bias, and confounding for each study using the following 6 questions adapted from Busse and Guyatt's Tool to Assess Risk of Bias in Cohort Studies [23]: 1) Was selection of exposed and nonexposed participants drawn from similar populations, 2) Was the follow-up adequate (e.g. percentage of attrition, percentage of missing data and type of missing data), 3) Can we be confident in the assessment of the exposure(s), 4) Can we be confident in the assessment of the outcome(s), 5) Did the study measure and adjust for potential confounding variables and covariates, 6) Can we be confident in the assessment of the confounding variables and covariates? The rating scale ranged from definitely yes (i.e. low risk of bias) to definitely no (i.e. high risk of bias). Rating disagreements between the 2 authors were resolved through discussion. As part of the assessment of potential confounding we also separately considered whether co-interventions were similar between the groups (i.e. social services).

Results
After removal of duplicates, 15940 records were identified. Of these, 1179 studies investigated the relationship between childhood adversities and biological health outcomes, of which 380 (32%) reported outcomes in children. A majority of the studies with outcomes in children used a cross-sectional study design (241 studies, 63%). We identified 97 longitudinal cohort studies in children, and 24 met our final criteria (see flow diagram in Fig. 1 for the exclusion criteria). In addition, we identified 11 studies that met all inclusion criteria from the references section of the selected studies, for a total of 35 studies (Fig. 1).

Bias assessment
The bias assessment synthesis is displayed in Fig. 2. In 77% of the studies, exposed and unexposed participants were drawn from similar populations, suggesting a low risk of bias. In regard to adequate follow-up, 68% of the studies were assessed to have a low or moderately low risk of bias. The assessments of the exposure and the outcome were rated as having a low or moderately low Table 1 Adverse Childhood Experiences (ACEs), age 0-19 years risk of bias in 94% of the studies. In regard to confounders and covariates, 92% of the studies received a rating of low or moderately low risk of bias (Fig. 2). In addition, we observed that the majority of the studies examining maltreatment and intimate partner violence did not take into consideration the possible effect of social services for abused and neglected individuals on the outcome.

Study characteristics
Of the 35 studies published between 2001 and 2015, 20 were published in 2010 and later. Thirty-three were prospective cohort studies and 2 were retrospective cohort studies. Half of the studies were conducted in the United States, and the remaining ones were conducted in 4 other countries. Sample size ranged from 18 to 13907 participants. Six studies examined developmental outcomes (Table 2) and 29 examined clinical outcomes (Table 3).

Developmental outcomes
Six studies investigated developmental outcomes including physical, reproductive, and cognitive development. Studies found that adversity was associated with differences in height, age at menarche, and cognitive ability (Table 3).

Physical and reproductive development
Studies on physical development showed a weak association between parental divorce or separation and shorter height, but the association was not always statistically significant. Li et al found that for males whose parents divorced or separated, height was significantly lower at ages 7 and 11 years, but not at age 16 years. For females whose parents divorced or separated, height was lower on average at ages 7, 11, and 16 years, but not significantly lower [24]. Li and Power compared 2 generations of participants and found that cohort members whose parents divorced or separated were shorter on average, and offspring were taller on average, but these associations were not statistically significant [25]. Boynton-Jarrett and Harville examined children who were raised in a neglectful environment (e.g. physical neglect, bullying) and found that 1 form of neglectful environment (OR = 1.21, 95% CI: 1.03-1.43) and 2 or more forms (OR = 1.76, 95% CI: 1.41-2.20) were both significantly associated with late menarche [26].

Cognitive development
The reviewed studies reported an association between early adversity and significant delays in cognitive development. Enlow et al found that interpersonal trauma exposure during age 0-24 months, but not during age 24-64 months, was significantly associated with decreased cognitive scores at 24 to 96 months of age [27]. Similarly, Strathearn et al found a significant association between maltreatment and cognitive delay at age 4 years [28]. Richards and Wadsworth found that children who experienced parental death or divorce had lower cognitive ability scores at ages 8 and 15 years [29].  Postnatal and recurrent maternal distress were significantly associated with higher cortisol levels, P = .007 and .006, respectively, but late distress was not, P = .112; the interaction between asthma diagnosis and recurrent distress was significantly associated with lower cortisol levels, P = .012, in children with asthma, whereas postnatal and late distress were not, P = .666 and .345, respectively; among children without asthma, cortisol levels were 46%, 26%, and 11% higher in children exposed to postnatal distress, recurrent distress, and late distress, respectively, P < .05; among children with asthma, cortisol levels were 33% higher and 13% lower in children exposed to postnatal distress and recurrent distress, respectively, P < .05 (the predicted change for late distress was near 0%) Essex  Somatic complaints at age 12 years (headaches, nausea, dizziness, tiredness, eye problems, aches, skin problems, stomach problems, vomiting, nightmares, and constipation) Adversity, particularly in the second 6 years of life, was associated with an increased risks of somatic complaints, but with differential effects by number of adversities and source of reporting: 5 or more adversities during the first 6 years of life were not significantly associated with child's report of somatic complaints, but were significantly associated with caregiver's report, OR = 3.31 (1.08-10.1), P = .04, and with either child's or caregiver's report, OR = 2.48 (1.05-5.87), P = .04; 5 or more adversities during the second 6 years of life were significantly associated with caregiver's report, OR = 3.37 (1.14-10.0), P = .03, but not with child's report of somatic complaints, OR = 0.70 (0.14-3.54), P = .67, or with either child's or   Husband's report of own and wife's aggression (HDV) at age 5 years was associated with a significantly smaller increase in baseline vagal tone from the first to the second assessment, P = .01, even when controlling for HDV at age 9 years, P = .01; wife's report of own and husband's aggression at age 5 years was not significantly associated with alterations in baseline vagal tone, P = .57 Schmeer (2012, USA) [44] Birth cohort from an urban area, followed to age 5 years (N = 1538) Parental separation or divorce between ages 3 and 5 years  Natural killer cell function at the fourth assessment and at all 4 assessments Parental psychiatric symptoms were not significantly associated with increased natural killer cell function at the fourth assessment, P = .13, but were significantly associated with higher natural killer cell function at all 4 assessments, P = .01 Studies were prospective cohort unless otherwise noted.

Clinical outcomes
Of the 29 selected studies investigating clinical outcomes, 14 focused on biological markers. Of these studies, 7 examined markers of endocrine function (cortisol), 4 examined markers of inflammation and immune function, 2 examined markers of autonomic nervous function, and 1 examined telomere length. Other clinical outcomes reported included obesity (6 studies), asthma (3 studies), infections and illnesses (3 studies), somatic complaints (3 studies), and sleep (3 studies).

Biological markers of endocrine function
Studies on cortisol, a stress hormone with a diurnal pattern of secretion, give insight to the dysregulation of the stress response in children exposed to adversity. Of the 7 selected studies that examined cortisol, 4 investigated the effect of maternal mental health on cortisol secretion and showed that results varied with age of exposure. Dreger et al found that maternal postnatal and recurrent distress were significantly associated with higher midday serum cortisol levels in 7-10-year-olds [30]. Essex et al assessed children's afternoon or evening salivary cortisol levels and found that those who were exposed to maternal depression both in infancy and at age 4.5 years had marginally significant, higher cortisol levels at age 4.5 years than did children never exposed or exposed only at age 4.5 years. In addition, children exposed to maternal depression only in infancy did not have significantly higher cortisol levels [31]. Halligan et al found that maternal postnatal depression was associated with higher, more variable morning salivary cortisol in 13-year-old children [32]. Ashman et al compared 7-8-year-old children of depressed and nondepressed mothers and found no effect of maternal depression on salivary cortisol levels or diurnal rhythm of cortisol [33]. In contrast, the 3 studies examining the effect of maltreatment on children's cortisol production reported lower cortisol levels in maltreated children. Peckins et al found that maltreated children were more likely than nonmaltreated children to have a blunted cortisol profile rather than a moderate or an elevated profile, but by late adolescence there was no longer a difference [34]. Two studies on cortisol response to a psychosocial stress test in 12-year-old twins indicated that maltreated or frequently bullied children did not exhibit the expected increase in cortisol following the stress test [35,36].

Biological markers of inflammation and immune function
The reviewed studies indicated alterations of immune and inflammatory response in children exposed to adversity. In a sample of children exposed to their parents' psychiatric symptoms, Wyman et al [37] and Caserta et al [38] found that higher levels of symptoms were associated with enhanced natural killer cell response in children, suggesting that chronic stress may exert effects of cytotoxicity on the developing immune system. Wolf et al found that greater parental depressive symptoms at baseline predicted increases in children's profiles of asthma-relevant inflammatory markers (i.e. eosinophil cationic protein and interleukin-4), in both children with asthma and controls [39]. In 3 cohorts of children, Copeland et al did not find an association between bullying or teasing and C-reactive protein levels (a marker of inflammation in the body) [40].

Biological markers of autonomic nervous function
Lynch et al found that the interaction between exposure to neighborhood crime, exposure to maltreatment, and children's genotype was associated with different patterns of respiratory sinus arrhythmia reactivity, a measure of the vagal tone indicating the physiological response of the autonomic nervous system to stress [41]. Rigterink et al found that exposure to domestic violence was associated with differences in the physiological regulation of emotion-measured as vagal tone-between exposed and nonexposed children, with smaller increases in baseline vagal tone in exposed children, suggesting a less adaptive development of regulatory functioning over time among these children [42].

Telomere length
One study investigated the relationship between violence exposure and telomere length, a marker of cellular age that also correlates with disease morbidity and mortality. Shalev et al observed stress-related accelerated telomere erosion in children who experienced 2 or more types of violence exposure, providing support for a mechanism linking cumulative childhood stress to potential lifelong and transgenerational health impacts [43].

Obesity
Evidence for early adversity influencing weight or body mass index (BMI) in childhood was mixed, with different types of adversity resulting in varying outcomes at different ages. Schmeer found that children whose mothers dissolved a union had an 80% higher risk of becoming overweight or obese between ages 3 and 5 years, as compared with children of stable married mothers [44]. Morris et al found that children whose parents had separated had a BMI 1.1% (95% CI: 0.2-2.0) higher at age 4 years in comparison with children whose parents remained together, but this diminished to 0.5% (95% CI: -1.3-2.2) by age 17 years [45]. Boynton-Jarrett et al found that children whose mothers reported chronic intimate partner violence (IPV) were more likely to be obese at age 5 years than were children whose mothers reported no IPV (OR = 1.80, 95% CI: 1.24-2.61), whereas children who were exposed to either early (up to age 12 months) or late (at age 3 and/or 5 years) maternal IPV did not have a significant increase in risk. Analyses stratified by sex indicated an increased risk for obesity only among females exposed to maternal chronic IPV (OR = 2.21, 95% CI: 1.30-3.75), as compared with males (OR = 1.66, 95% CI: 0.94-2.93) [46].
Studies on household dysfunction showed an effect on weight during early childhood, whereas studies on abuse and neglect reported an effect on weight during adolescence. Shenk et al found that maltreatment significantly increased the risk for obesity among female adolescents (RR = 1.47, 95% CI: 1.03-2.08, P = .034) [47]. Noll et al found a borderline, nonsignificant increase in risk for obesity during middle/late adolescence (age 15-19 years) for females who had been sexually abused (OR = 2.03, 95% CI: 0.54-4.60, P = .09); the relationship was not significant during childhood/early adolescence (age 6-14 years) [48].

Asthma
Studies reported that children exposed to childhood adversity had an increased risk for asthma. Lanier [51]. In addition, Kozyrskyj et al found that maternal longterm distress was significantly associated with an increased risk of asthma at age 7 years (OR = 1.25, 95% CI: 1.01-1.55), but postpartum distress and short-term distress were not [52].

Infections and illnesses
An increased risk for infection was also reported to be associated with childhood adversity. Lanier [38].

Somatic complaints
The relationship between childhood adversity and somatic complaints (i.e. common physical complaints of uncertain origin such as headaches and nausea) varied by number and types of adversities. Flaherty et al found that experiencing 5 or more adversities during childhood was not associated with child somatic complaints reported by the child at age 12 years, but was associated with child somatic complaints reported by the caregiver (OR = 4.26, 95% CI: 1.17-15.5, P = .03) [53].  [55].

Sleep
Childhood adversity was also shown to affect children's sleep. In a study on the effect of maternal depression on infant sleep, Armitage et al found that average sleep time in 24 hours did not differ between children of depressed and nondepressed mothers at 2 or 24 weeks, but nocturnal sleep was 97 min longer among children of nondepressed mothers at both 2 and 24 weeks, and these children had fewer awakenings during the night than did children of depressed mothers [56].

Discussion
To date, the majority of scientific inquiry on childhood adversity and biological health outcomes has focused on adult outcomes. Furthermore, most pediatric studies have been cross-sectional in nature, making it difficult to determine the temporal relationship between exposures to adversity and biological health outcomes. This systematic review synthesized the longitudinal evidence on childhood adversity and pediatric health outcomes to help provide insight into the link between early manifestations of a dysregulated stress response and biological health outcomes. Outcomes identified represent a range of conditions, which reflects the multiple systems impacted by a chronically dysregulated stress response in childhood. We observed great variability in the results of the longitudinal studies included in this review, suggesting that it is critical to consider both the nature of the adversity and the individual characteristics of the exposed child. The evidence from the selected longitudinal studies on cognitive delays, asthma, infections, somatic complaints, and sleep disruptions revealed associations between childhood adversity and increased risk of these conditions; this evidence adds to that from other nonlongitudinal studies in children [59][60][61][62][63][64][65][66][67][68]. This review also revealed mixed results related to obesity and physical development, with variability by age, sex, and exposure type. The variable findings may be due to bias related to loss to follow-up; however, genetic and environmental factors also may have influenced the heterogeneous response to childhood adversity [69][70][71][72][73][74]. Though a minority of the studies in this review focused on biomarkers, the growth of research and technology in this domain will help elucidate the biological mechanisms underlying the relationship between childhood adversity and health outcomes. Studies on cortisol included in this review used different sources (e.g. saliva, blood) and times of measurement, making it difficult to compare results across studies. In addition, the diverse effects of adversity observed on cortisol secretion may have been due to participant group differences and loss to follow-up. Nonetheless, the selected studies pointed to both elevated and blunted cortisol profiles, which might be indicative of a spectrum of responses that ranges from an initial increased activation of the stress response to a depletion of cortisol secretion over time, both of which can be detrimental to a child's developing body [75,76]. The reviewed studies also reported altered profiles of markers of inflammatory, immune, and autonomic nervous system functions, and epigenetic factors, highlighting the complexity of the biological response to adversity. It should be noted that, in general, the effect sizes reported across the various studies were small, which could be an accurate reflection of the small differences adversity exerts on the childhood outcomes we captured in this review or it could be an indication of methodological issues such as measurement or confounding.

Strengths and limitations
This systematic review's strengths include a wide publication window and a variety of search terms to capture studies on adversity. Limitations include exclusion of non-full-text and nonjournal publications, publication bias, and use of only 3 databases. We also did not contact authors for additional data or clarification. Another limitation is that definitions used for childhood adversity in the literature vary, making it challenging to perform a search to identify all sources of adversity. Furthermore, although most of the studies in this review examined more than one childhood adversity and some examined cumulative adversity, we did not explicitly focus on polyvictimization in our search, selection processes, or synthesis. Moreover, because the goal of this review was to broadly investigate relationships between a wide variety of adversities and health outcomes, we did not examine timing or severity of each individual type of adversity. Polyvictimization, timing, and severity are important aspects of childhood adversity [18,77], and future studies in these areas may give insight to the complex mechanisms underlying the dysregulation of the stress response.

Conclusions
This systematic review adds to the growing evidence on the relationship between childhood adversity and children's health. In particular, this review provides support for two important concepts: 1. Childhood adversity affects brain development and multiple body systems, and the physiologic manifestations are detectable in childhood. Although pediatricians can often recognize the behavioral signs associated with exposure to adversity [78], the evidence from this review suggests that pediatricians should also consider such exposure when evaluating the differential diagnosis of pediatric conditions such as developmental delay, asthma, somatic complaints, recurrent infections requiring hospitalization, and sleep disruption, which were found to be consistently associated with adversity in the selected studies. Although clinical diagnostic guidelines for detecting a toxic stress response in children do not yet exist, these conditions might provide part of the basis for the development of such guidelines. 2. The variability in children's response to adversity suggests complex underlying mechanisms, including the timing and severity of adversity, the experience of cumulative adversity, and the presence of protective factors (i.e. resilience, genetic and epigenetic constitution, individual socioeconomic status) that mitigate or exacerbate the impact of the exposure.
This variability poses a challenge for the development of uniform diagnostic guidelines [79,80].
Screening for exposure to childhood adversity, protective factors, and impacts of a toxic stress response could be the first step in providing targeted support for children at risk and their caregivers. In addition to screening, the American Academy of Pediatrics and other child health experts have also recommended strategies such as integrating behavioral healthcare into the pediatric home, offering parental support, providing peer-based education, and identifying community resources to help enhance resilience and mitigate the downstream effects of childhood adversity [81][82][83][84][85][86][87][88].
This systematic review provides a foundation for future research on the longitudinal relationship between childhood adversity and biological health outcomes. Future studies should examine in detail the findings we observed, taking into consideration timing and severity of adversity, the experience of cumulative adversity, and the presence of protective factors. This may be especially critical for outcomes for which we found mixed results, such as with physical development, obesity, and cortisol secretion. In addition, more large longitudinal studies are needed to develop a greater understanding of the causal pathways from exposure to childhood adversity to disease risk, to help the pediatric community develop services and interventions, identify protective factors, and prevent long-term negative health outcomes.