Deconstructing pathways to resilience: A systematic review of associations between psychosocial mechanisms and transdiagnostic adult mental health outcomes in the context of adverse childhood experiences

Abstract Adverse childhood experiences (ACEs) are identified with increased risk of adult mental health difficulties and negative impacts on well‐being. However, there is a need to go beyond simple associations and identify candidate mechanisms underpinning the ACEs–mental health relationship. Further methodological heterogeneity points to issues around the operationalization of ACEs and the importance of modelling data using robust research designs. The aim of the current review was to synthesize studies that utilized formal mediation and/or moderation analyses to explore psychological and social variables on the pathway between clearly defined ACEs (as measured by the ACE questionnaire and Childhood Trauma Questionnaire [CTQ]) and common mental health outcomes (depressive, anxiety and post‐traumatic stress disorder [PTSD] symptoms) across community samples aged over 18. A total of 31 papers were retrieved for critical appraisal. The majority of the studies explored factors mediating/moderating the link between child adversity and depression and less on anxiety and trauma. Most mechanisms were tested in only single studies, limiting the consistency of evidence. Evidence indicated that the mechanisms underlying associations between ACEs and adult mental health are likely to reflect multiple intervening variables. Further, there are substantial methodological limitations in the extant literature including the proliferation of causal inferences from cross‐sectional designs and both measurement and conceptual issues in operationalizing adversity. Consistent transdiagnostic mechanisms relevant to common mental health problems were identified, including perceived social support, emotion regulation and negative cognitive appraisals/beliefs. Further research using longitudinal design is required to delineate the potential contribution of the identified mechanisms.

However, a key conceptual issue in the current literature on ACEs is how researchers conceptualize and measure childhood adversity.
The definition of ACEs is relatively broad in that 'adversity' is used interchangeably with other concepts, such as 'trauma', 'traumatic experience', 'maltreatment' or 'stress'. The lack of clarity in the way 'adversity' is defined remains an important gap in the literature, spurring attempts to generate greater consistency in the operationalization of the term (McLaughlin, 2016). In addition, there are measurement challenges, with multiple questionnaires used for assessing ACEs (Zarse et al., 2019). For example, Aafjes-van Doorn et al. (2020) identified 127 questionnaires that have been used in ACEs research. Given the dose-response association between history of childhood adversity and the likelihood of later psychopathology (Dong et al., 2004;Merrick et al., 2017), methods of assessment that provide a cumulative score have been recommended (Bethell et al., 2017;Evans, Li, & Whipple, 2013). However, methodological and conceptual limitations of an additive ACE score remain (e.g., not taking into account the heterogeneity of adverse experiences) (Barboza, 2018).
Importantly, although ACEs have been conceptualized as a risk factor for psychopathology, there is no one to one mapping of ACEs to psychiatric disorder, ergo not everyone who has experienced childhood adversity will experience mental health problems (Belsky & Pluess, 2009). Evidence suggests the impact of ACEs may depend on various factors, such as the type of event and severity of exposure (Schalinski et al., 2016), the doseage or cumulative effects of adverse events (Edwards et al., 2003), age of exposure (Dunn et al., 2013;Riem et al., 2015, and the socio-economic context (Nurius et al., 2012). This suggests a need to understand the multi-factorial nature of ACEs and unpack mechanisms by which ACEs are implicated in potential trajectories of risk (psychopathology) and resilience.
Indeed, when exploring trajectories for psychopathology risk, recent transdiagnostic models have identified multiple shared underlying processes across various disorders, including genetic factors (e.g., Selzam et al., 2018), socio-environmental variables, such as poverty, family functioning and child maltreatment (Farah et al., 2006;Vachon et al., 2015), and psychological factors, such as emotion processing (McLaughlin, 2016). Furthermore, resilience-focussed research with youths and adults has shown that positive adaptation in the aftermath of adversity emerges from a dynamic interplay of various factors, including cognitive processes, coping strategies, individual traits, timing of the traumatic event, social context and support and relationship with caregivers (Masten, 2011;Rutter, 2006).
Transdiagnostic models also provide a novel framework for conceptualizing mental health problems accounting for multiple aspects, such as complexity, multifinality and comorbidity. Applying multifinality-by which a single risk factor can lead to a broad range of difficulties-to mental health has generated models of psychopathology that identify core and shared underlying processes across different types of psychopathology (Dalgleish et al., 2020;Nolen-Hoeksema & Watkins, 2011). There is also evidence that childhood adversity is associated with high prevalence of depression, anxiety and posttraumatic stress (e.g., Herzog & Schmahl, 2018;Lee et al., 2020;van der Feltz-Cornelis et al., 2019), while it has been argued that in the context of ACEs, these disorders may cluster together as a discrete group (Teicher & Samson, 2013). Similarly, an increasing body of research indicates comorbidity across depression, anxiety and PTSD (Forbes et al., 2011) which implies potentially cooccurring commonalities among these discrete disorders. Therefore, these findings highlight the need for further evidence as to how ACEs

Key Practitioner Message
• There is a need for early identification and implementation of primary, secondary and tertiary interventions with a view to targeting adversity at various levels.
• There is evidence for key mediating processes, such as social support, emotion regulation and cognitive processes, coping strategies, self-esteem and attachment, that could be taken into account in screening procedures, assessment and targeting of clinical or even preventative interventions.
• Encouraging social inclusion and support via group-based interventions could help improve symptoms of affective disorders in people exposed to child adversity.
• Therapists working with sufferers of abuse in the past could assess emotion regulation deficits and facilitate the development of emotion regulation skills, if needed, during treatment.
are related common mental health difficulties and their underlying mechanisms.
Despite the well-established link between ACEs and mental health, research exploring the mediational processes through which adversity leads to risk has been historically underdeveloped. However, over the last 10 years, there has been a growth in studies using formal mediation and moderation analysis techniques to explore the role of biological and psychosocial mechanisms in clinical and general populations. The aim of mediation is to strengthen quasi-causal inferences regarding the mechanisms through which an independent variable affects an outcome, while the aim of moderation is to examine variables affecting the strength/direction of the predictor-outcome relationship (Hayes & Preacher, 2010;MacKinnon, 2011). Therefore, mediation and moderation testing can help identify interdependencies between early adversity and mental health and explore the role of processes that lie between early adversity and mental health later in life. Importantly, examining third-variable effects is consistent with moving beyond associative relationships between variables towards identifying risk and protective factors and thus creating more stratified and nuanced models for mental health phenomenology and intervention.
Previous reviews have explored the link between adversity in childhood and psychopathology, focused on specific forms of adversity or traumatic experiences (e.g., Whiffen & MacIntosh, 2005), various types of psychopathology, including complex psychiatric disorders (e.g., Alameda et al., 2020), and specific mechanisms, such as neurobiological (e.g., McCrory et al., 2012) or cognitive ones (Aafjes-van Doorn et al., 2020), and included studies utilizing clinical or/and community samples. Hoppen and Chalder (2018) aimed to synthesize and critically evaluate the role of biological and psychosocial mediators and moderators on the link between child adversity and affective disorders in adults, including papers up to October 2017. This review was characterized reportedly by heterogeneity and utilized broad inclusion criteria, e.g., wide range of questionnaires measuring adversity, a variety of samples, such as clinical, non-clinical and specific populations, and inclusion of correlational as well as mediation and moderation studies, the latter forming a minority of the reviewed studies.
The current review sought to increase specificity through the adoption of narrower eligibility criteria, correspondingly reducing methodological variance. Specifically, to address the disparity that characterizes the literature base due to the operationalization and measurement of child adversity and improve specificity and validity, studies were only included if they measured child adversity via the ACE questionnaire (Felitti et al., 1998) or Childhood Trauma Questionnaire (CTQ) (Bernstein et al., 2003), which are the most widely used retrospective questionnaires of adversity in childhood that give a cumulative score of ACEs (Schmidt et al., 2020;Tonmyr et al., 2011;Zarse et al., 2019). Both questionnaires have been utilized extensively in research and were shown to have good psychometric properties (Bernstein et al., 2003;Dube et al., 2003). Furthermore, the search focused on three categories of mental health outcomes on the grounds that there is a high rate of comorbidity and common underlying psychological processes linked with anxiety, depression and PTSD in the context of ACEs (e.g., Purdon, 1999;Renna et al., 2017;Watkins, 2015). General population studies were included to overcome the effects of further confounding variables, prevent data skew and allow for generalisability of findings to the community.
Finally, studies were only included if they used formal approaches to mediation analyses, such as product of coefficient, difference in coefficient, Baron and Kenny approach, confirmatory test of complete mediation, and SEM or significance tests of mediation, such as Baron and Kenny's Causal-Steps approach (Baron & Kenny, 1986), joint significant test (Mackinnon et al., 2002), the Sobel first-order test (Sobel, 1982), PRODCLIN  and bootstrapping (Hayes, 2017;Johnson, 2001).
Therefore, the objective of the present review was to identify, synthesize and systematically evaluate studies that utilized formal mediation and/or moderation analyses to measure the effect of psychological variables upon the association between clearly defined ACEs (as measured by the ACE questionnaire and CTQ) and common mental health outcomes (depression, anxiety and PTSD symptoms) across general population samples.

| Literature search strategy
A narrative systematic review was carried out in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Moher et al., 2015). A systematic search of English language articles published from inception to September 2020 was conducted utilizing Medline, PsychINFO, Embase, CINAHL and ERIC. Google Scholar was also conducted to identify non-indexed papers. Identification of studies through manual review of reference lists of existing reviews in the field and retrieved articles was used to identify additional relevant articles. Keywords searched were related to the following three sets of terms combined with the Boolean operator 'and': (1) 'child adversity', (2) 'mediation' and 'moderation' and (3) terms representing mental health outcomes ('anxiety', 'depression' and 'PTSD/trauma symptoms'; see Appendix S1 for complete search strategy).
In the current literature, 'child adversity' is often used interchangeably with a wide range of terms (e.g., trauma and maltreatment) which were utilized as keywords to include a broad range of operational definitions of the term. Regarding the second set of key terms, both mediation and moderation analysis can help explore relationships between variables. First, mediation analysis is herein defined as the process that examines the effect of an intervening variable on the relationship between a predictor and an outcome variable (Fairchild & McDaniel, 2017). While several methods have been designed to investigate indirect effects between variables, there is not a unified approach in mediation with an agreed set of assumptions (Hayes, 2017). Second, moderation analysis is defined as the process that tests whether the prediction of the outcome from an Cross-sectional CTQ-SF; childhood adversity Self-concept clarity; Self-Concept Clarity Scale (Campbell et al., 1996) Abbreviations: ACE questionnaire, CDC-Kaiser Permanente Adverse Childhood Experience Questionnaire; Felitti et al., 1998); CTQ, Childhood Trauma Questionnaire (Bernstein & Fink, 1998;Bernstein et al., 2003); CTQ-SF, Childhood Trauma Questionnaire short form; DI, direct effect; IE, indirect effect. independent variable is different across different levels of a third variable (i.e., the moderator) Hayes, 2017).
Finally, the third set of key terms included three categories of mental health outcomes: depression, anxiety and PTSD or trauma symptoms. Studies were included if they used relevant mental health outcome measures.

| Eligibility criteria
Quantitative empirical studies were included in the review, if they (1) were written in English language and published as original research in a peer-reviewed journal, (2) used a cross-sectional or longitudinal research design, (3) reported exposure to ACEs before the age of 18, utilizing either the ACE questionnaire and adaptations (the original 7 item or its 10 item expanded form) or the CTQ questionnaire and its adaptations, (4) measured psychosocial (i.e., related to affect, behaviour, cognitions or mood as well as social and interpersonal processes) mediating and/or moderating factors, (5) used outcome measures for depression, anxiety and trauma/PTSD symptoms, (6) used formal methods for testing mediated or indirect effects (e.g., such as bootstrapping or SEM), and (7) used a general population sample aged over 18 years of age. There were no restrictions on study settings.

| Selection process and data extraction
After duplicates were removed, titles and abstracts obtained were reviewed against eligibility criteria. Full texts of studies meeting the search criteria were retrieved for further screening. Uncertainty regarding article eligibility was resolved through consensus discussion with the second author (AM). Data from studies meeting the inclusion criteria were extracted into a data extraction form to categorize relevant information from selected research articles (Tables 1 and 2).
Information was extracted from each individual study on (1) participant characteristics and sample size; (2) study design; (3) measurement of childhood adversity; (4) psychological and social mediating and moderating variables; (5) mental health outcome; (6) type of analysis; and (7) key research findings.

| Critical appraisal
Quality assessment was performed using a set of criteria which was adapted from a quality assessment tool previously utilized for critically evaluating mediation experimental studies (Cerin et al., 2009;Lubans et al., 2008;Mansell et al., 2013)] and subsequently modified to be utilized for observation studies (Lee et al., 2015). Additional criteria were added to the original tool to enable assessment of eligible research articles in terms of sampling procedures and bias, psychometric characteristics of independent variables, operationalization of mediators/moderators, reporting on the effect size, confidence intervals (CIs), and the variance accounted for in mediation models. The final assessment tool comprised 10 criteria (Table 6). Included studies were evaluated and given a score of 0 (weak), 1 (moderate) or 2 (strong) to each one of the criteria, while a total quality score was calculated for each study. A second rater critically appraised 30% of the included studies, with 83% agreement between raters. 3 | RESULTS

| Study selection
The initial search strategy identified 9,345 research studies out of which 5,481 were retrieved following removal of duplicates. Screening of reference lists of the eligible articles and searching Google Scholar yielded additional 17 articles. In total, 5,498 records were screened out of which 151 full texts were examined. Finally, 31 complete texts meeting the predefined search criteria were retrieved for critical appraisal. Studies were also excluded if they used exclusively clinical samples or high-risk community groups (e.g., veterans, male offenders, participants with substance misuse, etc.). See Figure 1 for study selection.

| Quality appraisal
The mean overall quality of the studies using community samples was moderate (63%), with a range from 45% to 90%. Mean overall quality for student studies was similar (57%), with a range from 30% to 80%.
The majority of studies were rated moderate to strong.
All included studies provided adequate evidence for their rationale behind the specific mediated and moderated effects they

Fitzgerald and
Gallus (2020) Significant indirect effect of child maltreatment on depression via emotional support from family and romantic partners but not friends. Hayward et al. (2020) Significant indirect effect of child adversity through self-concept clarity on depression. Both self-concept clarity and intolerance of uncertainty mediated the link between child adversity and mental health outcomes.
van Assche et al. (2020) Significant indirect effect of emotional neglect on depression through attachment anxiety. No direct effect found between emotional neglect and depression. Kogan et al. (2021) ACEs predicted young adult contextual stress which in turn forecast increases in defensive/hostile schemas. Defensive/ hostile schemas predicted increases in social developmental risk factors which were a proximal antecedent of depressive symptomatology and substance abuse no direct effect of ACEs on relational schemas reported. ACEs affect mental health outcomes indirectly through contextual contemporary factors. Nowalis et al. (2020) Anxious attachment to primary caregiver significantly moderated the child maltreatment-depression relationship. No significant effect identified for avoidant attachment to primary caregiver and for anxious or avoidant attachment to secondary caregiver. Song et al. (2020) Indirect effect of child maltreatment on depression though problem solving, self-blame, help-seeking, problem avoidance and rationalization. studies. A total of 23 self-report questionnaires were utilized for the measurement of outcomes, with the Beck Depression Inventory (BDI; Beck et al., 1996) most frequently used (n = 8).
A minority of studies (Kogan et al., 2021;Ross et al., 2019;van Assche et al., 2020;Wong et al., 2019) reported an a priori power analysis before data collection (fourth criterion). Proposed mediating variables were adequately defined by researchers in the majority of the studies, and valid measurements were utilized in most cases.
Therefore, the majority of the studies received strong ratings in this (fifth) criterion. Methods of data analysis employed were evaluated based on their power and statistical significance (sixth criterion).
Therefore, the causal steps approach was rated weak given its low power ; studies using the Sobel test (Sobel, 1982) in conjunction with the Baron and Kenny approach were evaluated as moderate, and finally, studies used estimation procedures for direct and indirect effects to test hypotheses regarding mediating effects (e.g., SEM and bootstrapping) (Agler & de Boeck, 2017;Hayes, 2009) were rated strong on this criterion. Of the total studies, 28 used SEM or process analyses with bootstrapping.
The majority of the studies reported effect sizes adequately and included CIs (seventh criterion).
Mediation models assume that mediators are affected by the exposure variable and that changes in mediators are associated with changes in the outcome. Since the majority of studies were crosssectional and data were gathered at a single time point, the temporal ordering of variables was speculative (eighth criterion).The majority of papers argued that longitudinal research would be required with a view to exploring how associations between variables play out with time. Approximately half of the included community-based and one third of the student-based studies reported the variance accounted for in their mediation models ) (ninth criterion).
Finally, the issue of potential confounding variables or covariates was considered in 6 out of 14 community-based studies and in 8 out of the 17 studies featuring students.

| Synthesis of psychosocial mechanisms
The majority of included studies (n = 25; 81%) explored factors that mediated or moderated the relationship between child adversity and depression; 12 studies (39%) investigated intervening variables in the relationship between child adversity and anxiety, and six studies (19%) examined mediators in the relationship between adversity in childhood and trauma. In some studies, the role of biological intervening variables was also tested (e.g., Kogan et al., 2021). These variables were outside the scope of the current review. The majority of the studies reported significant mediation effects. Key results of analyses per category of mental health outcomes are summarized below.

| Mediators/moderators of the relationship between childhood adversity and depressive symptoms
Twenty-four studies investigated the role of potential psychological mediators, and two tested social/interpersonal mediators. Of the studies that reported on putative psychosocial mediators in the relationship between child adversity and depression, 22 found that there was significant mediation, indirect effect and partial or complete mediation (Table 3).

Psychological mechanisms
Three studies examined the role of coping strategies as putative mediating variables of the childhood adversity-depression relationship. All identified emotion-oriented coping strategies as mediating the early adversity-depression relationship (Cantave et al., 2019;McQuaid et al., 2015;Song et al., 2020). Task-oriented coping strategies were found to function as a moderator in one study (Cantave et al., 2019). Song et al. (2020) tested the role of six coping strategies, via multiple mediation, reporting that problem solving, help-seeking behaviour, self-blame, problem avoidance and rationalization mediated the relationship between child maltreatment and depression. Of note, problem avoidance and rationalization did not mediate the effect among male participants. Self-blame was identified as the mechanism with the strongest mediating effect.
In a cross-sectional study (Crow et al., 2014), the authors hypoth-

McQuaid et al. (2015)
Iindirect effect of childhood trauma on depression through perceived discrimination. Partial mediation. Significant direct effect of childhood trauma on depression. Moderated mediation analyses showed that the intervening role of discrimination was stronger when levels of outgroup unsupport were higher. Ingroup unspport did not moderate the mediated relationship. Multiple mediation analyses: Emotion focused coping mediated the relationship between childhood trauma and depression. The path between emotion-focused coping and depression was moderated by both ingroup and outgroup unsupport. Wells et al. (2014) Significant mediational effect of child abuse on baseline depression scores through dysfunctional attitudes. Dysfunctional attitudes mediated the association of emotional maltreatment and depression. No significant effect was found for physical maltreatment alexithymia (an emotion regulation deficit) and found a significant indirect effect of emotional neglect on depression, indicating that alexithymia may be a potential mechanism linking early adversity (emotional neglect category) and depression.
Two studies (Hayward et al., 2020;Wong et al., 2019) reported that self-concept clarity may be a potential mediator linking early adversity and adult depression. Notably, when Wong et al. (2019) added self-esteem in the model, both the indirect effects through self-concept clarity and self-esteem were significant for depression, with the effect of self-esteem being larger. Additionally, Hayward et al. (2020) suggested that intolerance of uncertainty may also mediate the early adversity-depression relationship.

Social/interpersonal mechanisms
Three studies investigated social/interpersonal mechanisms mediating the early adversity and depression link. Using SEM analysis to test mediating effects of emotional support, Fitzgerald and Gallus (2020) identified support from family and romantic partners as mediators of the child maltreatment-depressive symptoms relationship. Consistent with previous research findings (Stafford et al., 2011), the results suggested that emotional support from friends did not have an intervening effect. Finally, Watt et al. (2020) investigated the effect of health behaviours and perceived social support in the relationship between adversity in childhood and depression in college students, concluding that perceived social support partially mediated the above relationship.

| Mediators/moderators in the relationship between childhood adversity and anxiety symptoms
Of the total studies, 12 (five community-based and seven with student-based) investigated the role of potential intervening variables in the child adversity-anxiety relationship. All studies reported on several psychosocial mediators of the relationship between child adversity and anxiety, revealing that there was an indirect effect, significant mediation or partial mediation (Table 4). Anxiety symptoms were measured with a variety of outcome measures, among which the General Anxiety Disorder Scale (GAD-7) was the most commonly used (33% of studies).

Psychological mechanisms
Self-concept clarity and intolerance of uncertainty were investigated as potential mediators on the path between ACEs and social anxiety, generalized anxiety and obsessive compulsive symptoms (Hayward et al., 2020). Hayward et al. (2020) reported that both self-concept Two studies investigated the role of repetitive negative thinking or rumination (Kim et al., 2017;Taylor et al., 2021). Research findings revealed that repetitive negative thinking mediated the early adversity-adult anxiety relationship. Kim et al. (2017) found that the effect was stronger among female participants. Taylor et al. (2021) also demonstrated that heightened attentional control moderated the indirect effect of negative repetitive thinking on the ACEs-anxiety relationship, indicating that it may be a risk factor for anxiety. Furthermore, two studies demonstrated that early maladaptive schemas and negative core beliefs mediated the association between maltreatment in childhood and adult anxiety (Berman et al., 2019;Gong & Chan, 2018).  Watt et al. (2020) found that perceived social support partially mediated the childhood adversity-anxiety relationship.

| Mediators/moderators of the relationship between childhood adversity and trauma symptoms
Of the total number of included studies, six investigated the role of potential intervening variables in the association between child adversity and trauma symptoms. They reported on five mediators and one moderator in the relationship between child adversity and trauma symptoms, all yielding significant effects (Table 5). Significant indirect effect of childhood adversity on anxiety via attachment anxiety-general. Gong and Chan (2018) Significant indirect effects of physical abuse, emotional abuse, and sexual abuse on anxiety through early maladaptive schemas. Reiser et al., 2014 Negative affect fully mediated the relationship between ACEs and health anxiety. Trait anxiety fully mediated the relationship between ACEs and health anxiety.

Psychological mechanisms
reporting that meaning making had an intervening effect on the child adversity-PTSD relationship.
Finally, a study by Wilson and Newins (2018)  Consistent with previous findings, the current review supports existing evidence for the link between early adversity and various forms of common psychopathology (e.g., Green et al., 2010;Hughes et al., 2017;Riedl et al., 2020) indicating that child adversity consti- With regard to the identified levels of scientific evidence, research findings found strong evidence for the role of both emotionoriented coping strategies and attachment anxiety upon the association between adversity in childhood and depression later in life. Furthermore, three high-quality studies explored the role of attachment avoidance, finding no evidence for mediation or/and moderation.
Therefore, there is strong evidence for the lack of effect of attachment avoidance upon the child adversity-mental health association.
Moderate evidence was also found for the intervening effect of emotion regulation, perceived social support (general, family, partner and friends), self-concept clarity and neuroticism in the relationship between adversity in childhood and depression. Neither strong nor moderate evidence for any of the examined variables was identified in studies that tested mediation and moderation models in the relationship between child adversity and trauma symptoms or anxiety. Overall, there is insufficient evidence for the majority of the putative mediating and moderating psychosocial variables as they were analysed in single studies.
On the basis of evidence from the papers included in the current review, we identify a number of candidate transdiagnostic mechanisms, particularly social support, cognitive appraisals and emotion regulation ( Figure 2). These mechanisms can be conceptualized as either risk or protective factors, operating across a spectrum of mental health difficulties, without disorder specificity for disparate disorders  (2007) Indirect effect of childhood trauma on trauma symptoms via cognitive distortions. Associations found between childhood trauma and attachment dimensions which on their own were not significantly associated with trauma symptoms. (Cludius et al., 2020;Compas et al., 2014;Conway et al., 2016). Other mechanisms explored in the reviewed studies were found to be relevant to anxiety and depression only (e.g., attachment anxiety and psychological inflexibility), while others were shown to be relevant to specific disorders (e.g., self-compassion and shame for depression and attentional control for anxiety). Of note, some seemingly disorderspecific mechanisms may be in fact transdiagnostic but are as yet untested (e.g., shame) or tested against a different measurement of adversity. From a lifespan perspective, the current findings are consistent with similar research in early adversity and psychopathology in adolescence, where the mediating role of processes identified in the current review, including emotion regulation, social support, cognitive appraisals and self-concept, has also been identified (

| Strengths and limitations
We acknowledge several limitations in the identified papers. The use of a large number of homogenous samples may limit the external validity of study findings, as could the use of non-random sampling (Banerjee & Chaudhury, 2010). The majority of the included studies also used volunteer selection introducing potential sampling bias or non-response bias (Cheung et al., 2017). Several studies used online recruitment platforms, such as Qualtrics and Amazon's Mechanical Turk, as they offer advantages over convenience sampling, such as student subject pools, by providing more diverse study participants.
Nevertheless, there is evidence that even this form of recruitment Further, in all studies, early negative experiences were measured retrospectively. It has been argued that this way of measuring past adversity may introduce potential measurement error into the findings, in the direction of either under-or over-reporting the impact of early adverse experiences on individual well-being (Colman et al., 2016;Hardt & Rutter, 2004). Of note, studies that compared retrospective and prospective methods of measuring ACEs varied between moderate (Reuben et al., 2016;Tajima et al., 2004) and low agreement (Baldwin et al., 2019). These findings should be interpreted in context and rather than indicating inadequate validity of retrospective measures, may instead reflect complementarity of both retrospective and prospective methods of measuring adversity (Newbury et al., 2018;Reuben et al., 2016). In addition, the use of cross- limitations, well-designed cross-sectional mediation studies can still give insight into pathways linking early adversity and adult mental health (Fairchild & McDaniel, 2017).
The systematic review also has several limitations. First, including only studies that utilized two well-validated measures of childhood adversity reduced variance, gave a clear, operationalization of adversity, and imposed a degree of homogeneity on a heterogeneous literature. However, equally, this was achieved at the cost of excluding studies measuring adversity using other measures of adversity exposure. Second, including only generalized non-clinical population was intended to increase specificity by reducing confounding variables emerging from specific or high-risk populations, but at the cost of impacting on generalizability to these same specific populations. Third, the current review included only published peer reviewed studies.
Fourth, we did not include research on biomarker mediators of child adversity and mental health, although there is some evidence of their role in affective disorders link (e.g., Koss & Gunnar, 2018). Finally, although we constrained the operationalization of childhood adversity, we did not do the same for outcomes. Consequently, there is likely measurement variance due different operationalizations of mental health outcomes. given developmental period (e.g., childhood). Furthermore, the questionnaires utilized in the included studies predominantly focus on relational types of adverse experiences that themselves largely occur within the familial environment. However, there are diverse types of adverse experiences that could also be the focus of future research, such as violence in the community, bullying at school, racial discrimination, medical crisis and war (Cross et al., 2015;Halevi et al., 2017;Strasshofer et al., 2018).

| Implication for research, practice and theory development
The current review focused on anxiety, depression and PTSD/ trauma on the basis of the potential for common underlying transdiagnostic processes (Gardner et al., 2019;Teicher & Samson, 2013).
Given the number and breadth of studies on ACEs and common mental health outcomes, meta-analytic studies of potential mediating variables in the relationship between child adversity and mental health using both clinical and community samples could be useful. However, there are methodological challenges herein due to the heterogeneity of potential psychosocial mechanisms, outcome measures, mediation analysis approaches and hence, methods of reporting results, merit consideration.
With regard to clinical implications, we highlight the need for early identification and implementation of primary, secondary and tertiary interventions with a view to targeting adversity at various levels (Thoresen & Olff, 2016). There is evidence for key mediating processes, such as social support, emotion regulation, and cognitive processes, coping strategies, self-esteem and attachment, that could be F I G U R E 2 Mechanisms linking childhood adversity and psychopathology taken into account in screening procedures, assessment, and design of clinical or preventative interventions. Therefore facilitating improvement in these psychological targets could produce corresponding improvement in mental health symptoms. Specifically, these could be targeted and used as an adjunct to traditional approaches to treating depression, anxiety and PTSD in populations with experiences of adversity in childhood (e.g., cognitive-behavioural therapy [CBT]; Cuijpers et al., 2016) or form the basis of stand-alone interventions.
On the basis of the current findings, targeting social inclusion and support via group-based interventions could also impact on affective symptoms in people with a history of adverse experiences in childhood. Furthermore, since emotion dysregulation appears to be one of the mechanisms that could explain the child adversitypsychopathology association (Crow et al., 2014;Klumparendt et al., 2019;Stevens et al., 2013), therapists working with sufferers of abuse in the past should remain mindful to possible emotion regulation deficits and facilitate the development of emotion regulation skills, if needed, during treatment. There is preliminary evidence that participation in programmes designed to facilitate development of emotion regulation skills in people with experiences of adversity can promote people's resilience and psychological well-being (Cameron et al., 2018). Finally, digital or peer-led interventions and/ or preventive programmes in the community could encourage a greater use of task-oriented strategies to help tackle the maladaptive use of emotion-oriented coping strategies shown to explain the child adversity-affective disorders link (Cantave et al., 2019;McQuaid et al., 2015).
The use of non-clinical community-based samples in the review improves generalizability to the community and, as such, can help inform public mental health approaches and the design of preventative programmes. From a preventative perspective, our findings highlight the role of coping as a mediator, suggesting that a more general focus on improving children's self-esteem, self-concept clarity, taskfocused coping strategies and teaching adaptive ways to express and regulate their emotions could be targeted as part of early intervention and prevention programmes designed to be delivered by mental health professionals in various settings, including schools.
Furthermore, children and adolescents who have suffered adverse experiences within their familial environment often do not seek support early, leading to help-seeking difficulties and a long duration of under-treated symptoms, further supporting the impetus for early intervention and the development of preventative programmes.

| Conclusion
Overall, our review highlights the need to prevent ACEs and develop interventions aiming to provide support to individuals exposed to early adversity. The wide range of potential mediators examined in the included studies foregrounds the difficulties in unifying these factors within a single theoretical framework. Indeed, the association between child adversity and psychopathology is multifactorial and context-related. We emphasize the importance of further research on the field with a view to garnering more evidence-based information about factors that influence the link between childhood adversity and specific mental health outcomes. Finally, another crucial issue is the lack of consensus criteria upon the way adversity and trauma are conceptualized. These criteria should be established and agreed upon with a view to supporting intervention, research and policy development.