Study Inclusion

The results of inclusion and exclusion at various phases are described in a PRISMA chart (see Fig. 2; Page et al., 2021). Studies were excluded for not meeting one (or more) of the inclusion criteria. Overall, the search yielded 47,241 records. The following number of records were retained respectively after de-duplicating, title screening, abstract screening, and full-text screening: 33,536, 4,805, 736, and 216. The full texts of the remaining 216 records were closely read, with data extracted from 140 records and 76 papers further excluded. The 140 records fell into two distinct groups in terms of peer relations: (1) general peer relationships (n = 92); and (2) bullying (n = 51). This paper reports the 92 records on general peer relationships. The records on bullying are reported in Merrin et al. (2023). Several of the 92 studies were based on the same large public datasets (e.g., National Longitudinal Study of Adolescent to Adult Health [Add Health], National Survey of Child and Adolescent Well-Being [NSCAW], Longitudinal Studies of Child Abuse and Neglect [LONGSCAN]; see Supplemental Appendix F for more details).

Fig. 2

PRISMA flowchart. aFour of the articles examined models that included bullying and general peer relationship (PR) variables in the same model, with bullying as the dependent variable and general PR as either a mediator or moderator. bReferences in this group examined separate models for bullying and general PR variables in relation to ACEs. As such, these three references are reported in both this manuscript and Merrin et al. (2023)

Study Characteristics

An overview of the study characteristics of the 92 included studies is provided in Table 1 (see Supplemental Table S1 for more information). There was an increase in publications in recent years (e.g., 57% were published in 2014 or later; see Fig. 3). Most studies were conducted in the US (n = 50). Studies were also conducted in other countries, including Canada (n = 7), South Korea (n = 5), the UK (n = 5), Sweden (n = 4), Brazil (n = 2), Finland (n = 2), Germany (n = 2), Israel (n = 2), and Norway (n = 2). One study was conducted in each of the following countries or regions: Bosnia and Herzegovina, Denmark, France, Haiti, Hong Kong, India, Ireland, New Zealand, Spain, and The Netherlands. Lastly, one study was conducted across four European countries (Austria, Germany, Slovenia, and Spain). Across the 11 singular and five combined ACEs (see Fig. 4), sexual abuse was most studied in relation to peer relations (n = 21), followed by physical abuse (n = 16), divorce and separation (n = 14), and child welfare involvement (n = 14). The other ACEs all had fewer than 10 studies.

Table 1 Overview of included studiesFig. 3

Number of publications per year included in this review. The number of publications for 2019, 2020, and 2021 is an underestimate of the actual publications that would have met all inclusion criteria because the search was implemented in September and October of 2019. There are some publications for 2020 and 2021 because some of the publications (i.e., advance online publications) were updated with volume (issue) and page numbers and publication years by the journals after the search had been implemented. For example, Yoon et al. (2021) was published online in 2018 and was included in the search

Fig. 4

Number of studies per ACE that examined ACE in relation to peer relationships. ACE adverse childhood experience. ACEs in uppercase indicate studies that analyzed combined ACEs (i.e., a score of more than one singular ACE) in relation to peer relationships. For instance, LeBoeuf et al. (2017) examined girls’ abuse history (either physical or emotional), thus was counted in the ABUSE group. Some studies analyzed more than one ACE score, thus were counted more than once. For example, Lim and Lee (2017) analyzed abuse and neglect as two separate ACEs, thus was counted in both the ABUSE and NEGLECT groups. The numbers in this figure include studies on the associations between ACEs and peer relationships, as well as the mediating, moderating, and risk and protective effects of peer relationships

The total score of the quality assessment ranged from 0.50 to 0.95 across studies, with an average of 0.78. Some studies received a high score due to their high quality. Others received a high score because their designs were simple, and several assessment items were not applicable. Readers are encouraged to examine the quality score at the item level (see Table 2) and consider the quality score in combination with the study design. The scores for items ranged from 0.89 (outcome measure) to 1.99 (study objective).

Table 2 Quality assessment scores of included studies at the item levelStudy Results

Results are described according to the six aspects of peer relationships. In each aspect, analyses are reported in the following sequence (when relevant): (1) associations between ACEs and peer relation variables (i.e., main effect); (2) peer relations as a mediator in the associations between ACEs and youth outcomes (or indirect effect for cross-sectional studies); (3) peer relations as a moderator of the associations between ACEs and youth outcomes; and (4) risk and protective effects of peer relationships. The first three types of analyses included studies that had participants with and without ACEs and treated ACEs as independent variables. The last type of analysis was based on samples (or subsamples), all of whom had the same ACEs, and treated ACEs as a sample inclusion criterion.

Youth outcomes in the mediation and moderation models, as well as the risk and protective effects section, included any outcome variables, such as academic achievement, mental health, and aggression. Within the main effects section, the results are sorted by ACEs, starting with combined ACEs (e.g., cumulative ACE score) and then singular ACEs (e.g., sexual abuse) when relevant. The results are presented in a separate subsection if there are two or more studies on the relation between a specific ACE and peer relations. ACEs with only one study are grouped in one subsection.

Among the mediation models, the paths between the independent variables (i.e., ACEs) and mediators (i.e., peer relations) are also described in the main effect section. Studies on the moderating effect of peer relations rarely reported relations between ACEs and peers. However, when such associations were reported, they were also described in the main effect section. There are two caveats concerning the mediation. First, some studies examined the mediating effect of peer relations in cross-sectional data with no temporal order, which technically are indirect effects (not mediation). Second, some studies reported significant paths between independent variables and mediators, and between mediators and dependent variables, but did not report a formal test of the indirect effect. These studies are described as “possible mediation.”

Peer Relationship QuantityAssociations Between ACEs and Peer Relationship Quantity

Overall, negative relations between ACEs and peer relationship quantity were reported in four of the six studies (see Table 3) utilizing various analysis strategies (i.e., ANOVA, linear regression, regression with propensity score matching), although nonsignificant findings were reported in several of these studies depending on the variable or comparisons (e.g., nominations sent vs. received; Bryan, 2017). The remaining two studies reported either nonsignificant (Wolfe, 2016) or positive associations (Seeds et al., 2010). However, the magnitude of the differences in peer relationship quantity between youth with and without ACEs was small, regardless if it was a negative or positive association (e.g., 0.4–0.5 fewer friends, Bryan, 2017; 1.13 more friends, Negriff et al., 2015). Across all ACEs, parental incarceration was negatively associated with peer relation quantity.

Table 3 Number of studies regarding the associations between ACEs and peer relationship aspects

Peer relation quantity was assessed via peer nomination and self-report. An upper limit was set in the peer nomination studies (i.e., up to five male and five female friends). The self-report method included: (1) asking the number of friends (single-item); (2) asking the names of friends and counting the total of the names; or (3) asking if the number of friends was within 0–2 or not. All analyses were cross-sectional. Samples ranged from small convenient (e.g., N = 101; Seeds et al., 2010) to large national samples (e.g., N = 2,575,269; Fontes et al., 2017).

Maltreatment

The associations between maltreatment and peer relation quantity were mixed (i.e., negative, positive, or nonsignificant) across the two studies. Negriff et al. (2015) found that youth in the maltreatment group reported having fewer same-aged friends than the comparison group (​​4.43 vs. 5.56, p = .01) at Time 2 (10–13 years) but not Time 3 (11–14 years). However, no significant differences were found in the number of older friends at Time 2 and 3 (Negriff et al., 2015). Seeds et al. (2010) reported youth with maternal maltreatment had more close friends (5.00 vs. 3.38, p < 0.05) than counterparts who did not have maternal maltreatment. This association was further demonstrated in a path analysis. This was the only study that reported a positive association between ACE and peer relationship quantity. The authors speculated that adolescents from abusive families might rely more on peers for support (Seeds et al., 2010). Nevertheless, paternal maltreatment was not associated with the number of close friends (Seeds et al., 2010). Of note, the maltreatment in Negriff et al. (2015) did not include sexual abuse, but Seeds et al. (2010) did. The mixed findings across the two studies also suggest relations between maltreatment and peer relation quantity may be subject to several factors’ influence, including youth’s age, same-aged versus older friends, and maternal versus paternal maltreatment.

Parental Incarceration

Negative relations between parental incarceration and peer relation quantity were reported in two studies, both of which analyzed Add Health data. Utilizing propensity score matching and regression, Cochran et al. (2018) reported negative associations of parental incarceration with friendship nominations received and sent in a less conservative matching model. However, such associations became nonsignificant in a more conservative matching model. Thus, differences in friendship nominations may have been an artifact of the matching strategy. Using regression analysis, Bryan (2017) reported that father incarceration was associated with fewer friendship nominations sent but not received (i.e., these adolescents were more likely to withdraw from peers rather than be excluded). However, the magnitude of the difference in friendship nominations sent (i.e., 0.4–0.5 fewer friends) was small in an absolute sense and may not carry substantively meaningful implications for practice.

Other ACEs

One study examined sexual abuse, and one investigated caregiver substance misuse in relation to peer relationship quantity. In a very large cross-sectional Brazilian study (N = ​​2,575,269) with propensity score matching and regression, Fontes et al. (2017) found that youth who were sexually abused were more likely to report having a limited number of friends (i.e., 0–2 friends) compared to peers who had similar characteristics but were not sexually abused. In a cross-sectional US national sample (N = 4,147), Wolfe (2016) reported no significant difference in the number of self-reported friends (16.80 vs. 16.49, ns) between youth whose mothers had alcohol use disorders and those whose mothers did not.

Quantity as a Mediator

In a cross-sectional study, Seeds et al. (2010) found that maternal maltreatment was positively associated with the number of close friends, which was further negatively related to depression severity. However, the indirect link was nonsignificant.

Summary

ACEs seemed negatively associated with peer relationship quantity. This association may be subject to several factors, including adolescents’ age, age of friends, friendship nominations received versus sent, and from whom the ACEs originated (e.g., maternal vs. paternal). However, the literature was small, and multiple ACEs were not studied in relation to peer relationship quantity. Lastly, it is unclear how substantively meaningful it might be for youth with ACEs to have a small but statistically significant difference in friendship quantity.

Peer Relationship QualityAssociations Between ACEs and Peer Relationship Quality

Across all ACEs (see Table 3), caregiver substance misuse and household mental illness were negatively associated with peer relation quality; maltreatment was not significantly related to quality. The associations between the other ACEs and peer relation quality were negative and nonsignificant, except Lepistö et al. (2012), who reported that combined physical and emotional abuse was associated with a higher likelihood of reporting a mutual friendship. Such results were reached through diverse sample sizes (i.e., ranging from small [e.g., N = 88; Shapiro & Levendosky, 1999] to large [N = 164,580; Bergström et al., 2013]) and various analyses (i.e., chi-square test, t-test, ANOVA, ANCOVA, logistic regression, linear regression, regression with propensity matching, path analysis, and SEM). However, most studies were cross-sectional, and peer relationship quality was primarily measured via self-report, except for three studies that utilized parent-reports (Katz et al., 2007; Morón-Nozaleda et al., 2017; Roetman et al., 2019). The quality of measures ranged from single-item (e.g., Wolfe, 2016) to multi-item measures with sound psychometric properties (e.g., Farruggia et al., 2006).

Maltreatment

Maltreatment was not significantly related to conflicts with friends (Shapiro & Levendosky, 1999; N = 88) or best friend satisfaction (Levendosky et al., 2002; N = 111). Of note, the sample sizes of both studies were relatively small.

Abuse

Mixed (i.e., negative, positive, or nonsignificant) findings were reported on the associations between abuse and peer relation quality across four studies. Combined physical and emotional abuse had a negative association with peer relationship quality (Ban & Oh, 2016) and peer attachment (Ju & Lee, 2018), but was associated with a higher likelihood of youth reporting having mutual friendships (Lepistö et al., 2012). Lepistö et al. (2012) was the only study that reported a positive association between ACE and peer relationship quality (i.e., mutual friendships). Youth who were abused may rely more on friends than their home relationships compared to their counterparts who did not experience abuse at home (Lepistö et al., 2012). Lastly, abuse (no specification of forms) was not related to peer attachment (Lim & Lee, 2017).

Neglect

Negative and nonsignificant findings were reported regarding the relations between neglect and peer relationship quality. Specifically, neglect was negatively related to general peer relationship quality (Ban & Oh, 2016; N = 2070) and peer attachment (Lim & Lee, 2017; N = 2351) but was not associated with peer relation maladjustment (i.e., dysfunctional peer relations; Kwak et al., 2018; N = 1170). Although all three were cross-sectional studies conducted in South Korea, the sample sizes and variables controlled in the analyses differed.

Sexual Abuse

Sexual abuse was negatively associated with friendship quality in Greger et al., (2016; N = 1352) but not related to conflict with friends in Shapiro and Levendosky (1999; N = 88). The relatively small sample size of Shapiro and Levendosky (1999) should be noted.

Family and Domestic Violence

Negative associations of family and domestic violence with peer relationship quality were reported in two studies, and one study reported nonsignificant findings. In a cross-sectional study across four European countries, Kassis et al. (2018) reported youth who were “resilient” to family violence (i.e., no depression or aggression despite experiencing family violence; n = 510) expressed lower levels of empathy toward peers than the comparison group (i.e., no family violence, no depression or aggression; n = 2055). In a longitudinal study, Narayan et al., (2014; N = 182) reported domestic violence exposure before age 5 was positively related to conflicts with best friends at age 16. In another longitudinal study, Katz et al., (2007, N = 65) reported a nonsignificant association of domestic violence exposure at age 5 with negative peer interactions or friendship closeness at age 11. The relatively small sample size might explain the nonsignificant association. Also, Katz et al. (2007) used mother-reported peer relation quality, whereas Kassis et al. (2018) and Narayan et al. (2014) used youth self-reports. Lastly, the definition of peer relationship quality differed across the three studies.

Caregiver Substance Misuse

Caregiver substance misuse was associated with negative peer relationship quality (i.e., higher peer stress; “not close to friends”) in two studies. Specifically, Marshal et al. (2007) found parental alcoholism was related to higher peer stress. Wolfe (2016) reported maternal alcohol use disorder was positively associated with youth reporting “not close to friends,” although not related to “difficulty to make friends.”

Household Mental Illness

Household mental illness was related to negative peer relationship quality in two studies, with differences between fathers’ and mothers’ mental illnesses. In a cross-sectional study conducted in Spain, Morón-Nozaleda et al. (2017) found youth whose parents had bipolar disorders reported more peer relationship problems than the community control group. In a longitudinal study in Sweden, Roetman et al. (2019) reported maternal mental disorder (including substance disorder) before the child’s 10th birthday was associated with higher odds of peer relation problems at age 15. However, paternal mental disorder was not associated with peer relationship problems.

Divorce and Separation

Mixed (i.e., negative or nonsignificant) findings were reported regarding the associations of parental divorce and separation with peer relations quality across five studies. All studies examined family structure (e.g., stepfamily, single-parent family, two-parent family), which indicated possible divorce and separation. In a UK study (N = 1515), Attar-Schwartz et al. (2009) reported that living in stepfamilies was positively associated with peer relation problems compared to two-parent families and that single-parent families were not associated with peer problems. In a Swedish study (N = 164,580), Bergström et al. (2013) reported that joint child custody and living with one parent were associated with lower peer relationship quality at age 12 compared to two-parent households. However, at age 15, joint child custody was no longer associated with peer quality, although living with one parent was still associated with lower peer relation quality. Findings highlighted additional factors to consider, such as the youth’s age when the divorce or separation occurred.

The three other studies reported nonsignificant associations of family structures with peer relationship quality, intimacy, admiration, as well as conflicts and problems with peers (Flouri & Panourgia, 2011, N = 430, UK; Noack et al., 2001, N = 637, Germany; Tarter et al., 2001, N = 91, US). The sample sizes of these three studies were much smaller than the two prior studies. Studies examining divorce and separation might require a larger sample size, as the total sample is often divided into more than two groups to reflect multiple family structures and the non-nuclear family structure groups usually represent a small percentage of the total sample. Also, the effect of family structure on peer relation quality might be small, although statistically significant. Lastly, these five studies were from five different countries.

Child Welfare Involvement

Negative associations between child welfare involvement and peer relation quality were reported in two studies, and one study reported nonsignificant results. Youth in foster care had lower friendship quality than youth in the general population, as rated by themselves and their primary caregivers (Jozefiak & Sønnichsen Kayed, 2015). Youth in foster care were also less likely to report “friends care a lot” (Perry, 2006) than youth in the general population, but did not differ from their counterparts in terms of perceived peer warmth (Farruggia et al., 2006). The definitions of peer relation quality differed slightly across the three studies, yet they all compared peer relation quality of youth in foster care with their counterparts.

Other ACEs

One study examined the associations of the following ACEs with peer relationship quality: the cumulative ACE score, household dysfunction, physical abuse, and parental incarceration. The cumulative ACE score was negatively associated with peer relation quality in a short-term longitudinal study (Landers et al., 2020). A nonsignificant association between household dysfunction (i.e., parental psychopathology, criminality, and alcohol or substance abuse) and peer relation quality was reported in a cross-sectional study (Greger et al., 2016; N = 1352). Physical abuse was not associated with friendship quality (Greger et al., 2016). Lastly, Cochran et al. (2018) reported in a nationally representative sample that parental incarceration was not associated with youth reporting having a best friend.

Quality as a Mediator

Mixed (i.e., significant or nonsignificant) findings were reported regarding the mediating role of peer relation quality across seven studies. When a mediating effect was found, youth with ACEs tended to have lower peer relation quality which in turn had negative implications for their well-being. A strength of the literature was that most studies provided a formal significance test of the mediation, except Kwak et al., 2018 (this was irrelevant due to the nonsignificant path between the independent variable and mediator). However, most studies included fewer than three waves of data except for Ju and Lee (2018) and Narayan et al. (2014). The studies included five ACEs as independent variables, six variations of peer relation quality as mediators, and nine youth outcomes as independent variables.

Specifically, Landers et al. (2020) reported that positive peer relationships mediated the association of the cumulative ACE score with youth behavioral needs (e.g., conduct problems, oppositional behavior, and anxiety) and life functioning (e.g., social functioning, employment, and living situation). Ban and Oh (2016) found general peer relation quality mediated the associations of abuse and neglect with emotional and behavioral problems. Further, peer attachment mediated the relations of abuse with depression (Ju & Lee, 2018), as well as the relations of neglect with school adjustment (Lim & Lee, 2017). Lastly, peer stress mediated the relations between parental alcoholism and youth alcohol involvement in participants with ADHD but not in participants without ADHD (Marshal et al., 2007). Of note, the interaction between parental alcoholism and ADHD (in relation to peer stress) was “significant” at 0.15 in Marshal et al. (2007), indicating it was not advisable to proceed with mediation analysis within the two subsamples (ADHD vs. comparison).

Nonsignificant findings were also reported. Specifically, positive peer relationships did not mediate the association of the cumulative ACE score with youth risk behavior (Landers et al., 2020). Peer attachment did not mediate the relation between abuse and school adjustment (Lim & Lee, 2017). Further, peer relation maladjustment did not mediate the relation between neglect and smartphone addiction (Kwak et al., 2018). Lastly, conflicts with best friends at age 16 did not mediate the relations between domestic violence exposure before age 5 and young adult dating violence (Narayan et al., 2014).

Quality as a Moderator

Both significant and nonsignificant moderating effects of peer relation quality were reported across the five studies. Specifically, problem behaviors with best friends amplified the association of physical abuse with youth delinquency (Salzinger et al., 2007). Positive peer relationships mitigated the negative relation between maternal alcohol use disorder and youth mental health (Wolfe, 2016). However, the time youth spent with friends did not moderate the relation between the cumulative maltreatment score and youth offending behavior (Wilkinson et al., 2019). Peer attachment did not moderate the relation between physical abuse and youth delinquency (Salzinger et al., 2007). Peer closeness did not moderate the association between family violence and later youth anxiety (Goodearl et al., 2014). Lastly, peer connectedness did not moderate the relation between physical abuse and nonproblematic behavior (Merritt & Snyder, 2015).

Risk and Protective Effects of Quality

Overall, the seven studies indicated that high quality peer relationships were a protective factor for youth with ACEs, whereas low quality was a risk factor, with caveats to consider (e.g., internalizing vs. externalizing problems; S. Yoon et al., 2021). Such results were reached through various analyses (i.e., t-test, linear and logistic regression, path analysis, and latent growth models) and diverse sample sizes (i.e., ranging from 73 [Jaser et al., 2007] to 1,054 [Merritt & Snyder, 2015]). However, most of the studies were cross-sectional except for S. Yoon et al. (2021).

Six of the seven studies examined child welfare involvement as an ACE, three of which utilized NSCAW-II or NSCAW data. Utilizing longitudinal NSCAW-II data, S. Yoon et al. (2021) reported in a sample of youth investigated by child protective services that peer relation satisfaction was associated with low initial levels of and a slower rate of decline in internalizing and posttraumatic stress symptoms across three waves. However, peer relation satisfaction was not related to externalizing behaviors. Utilizing the same dataset (NSCAW-II), Merritt and Snyder (2015) reported that perceived school peer connectedness was associated with increased odds of nonproblematic behavior (i.e., T score lower than 60 on the Child Behavior Checklist) among youth in child welfare. Lastly, H. M. Thompson et al. (2016) reported in NSCAW data that low peer relation quality was related to more externalizing behaviors, internalizing behaviors, and delinquency, as well as a lack of self-esteem among youth in foster care.

In addition, Farruggia et al. (2006) reported among youth in foster care that peer warmth was positively related to concurrent youth mental health but not achievement or misconduct. J. W. Kim et al. (2015) conducted a cross-sectional t-test in a South Korean child welfare-involved sample and found youth with low quality peer relations reported lower cognitive functioning and more internet addiction, ADHD, depression, anxiety, and suicidal ideation. Perry (2006) reported among youth in foster care that peer network strength was positively related to foster network strength, but was not associated with youth depression, anxiety, or biological network strength.

Lastly, in a cross-sectional sample of 73 adolescents, all of whom had a parent with depression, Jaser et al. (2007) found that primary control coping (e.g., problem-solving) of peer stress was negatively associated with self-reported anxiety, depression, and aggression, but not parent-reported aggression. Secondary control coping (e.g., distraction) was negatively associated with anxiety and depression, but not aggression (neither self- nor parent-report). Disengagement coping (e.g., avoidance) was not related to anxiety, depression, or aggression.

Summary

Peer relation quality plays a critical role in the well-being of youth with ACEs, as evidenced in the mediating, moderating, and risk and protective effects. However, not all ACEs were included in the examination of these effects. In particular, the risk and protective effects were investigated in samples mostly consisting of youth in child welfare. The associations between ACEs and peer relation quality were studied in more ACEs, three of which showed consistent patterns: Caregiver substance misuse and household mental illness were negatively related to peer relation quality, whereas maltreatment was not significantly associated with quality. The associations between the other ACEs and peer relation quality were mostly a mix of negative and nonsignificant findings. Such associations may be subjected to multiple factors, including participant age when the ACEs occurred (Bergström et al., 2013), which caregiver the ACEs were from (e.g., maternal vs. paternal; Roetman et al., 2019), and how peer relation quality was defined (e.g., general quality, intimacy, and conflicts with friends). Lastly, the presence of protective factors (e.g., closeness) was rarely considered alongside the absence of risk factors (e.g., conflicts with friends) of peer relation quality in relation to ACEs.

Peer CharacteristicsAssociations Between ACEs and Peer Characteristics

Across all ACEs examined (see Table 3), emotional abuse, caregiver substance misuse, and parental incarceration were associated with negative peer characteristics. Composite ACEs were not significantly related to peer characteristics. The relations of the other ACEs with peer characteristics were a mix of negative or nonsignificant findings, except that Negriff and Trickett (2012) found maltreatment was related to positive peer characteristics (i.e., lower levels of peer alcohol use). Such results were reached through a range of ana