Childhood emotional maltreatment and trauma-related outcomes: a systematic review and multilevel meta-analysis

ABSTRACT

Background: Estimating the association between childhood emotional maltreatment (CEM) and trauma-related outcomes is important for prevention and intervention.

Objective: We synthesised the evidence by differentiating between emotional abuse and emotional neglect, examining post-traumatic stress disorder (PTSD) and complex PTSD (CPTSD) and exploring moderators and mediators.

Method: Following PRISMA guidelines, seven databases were searched up to 19/03/2026 (PROSPERO-CRD42024598358) for studies on the association between CEM and PTSD/CPTSD. Study quality was assessed using an adapted Newcastle-Ottawa Scale (NOS), and certainty of evidence was evaluated using the GRADE framework.

Results: Of 11,479 records, 62 were included in the review, of which 57 studies provided 132 effect sizes for a random-effects multilevel meta-analysis (N = 323,776). Study quality was overall fair (Mean NOS = 5.8). A positive association between CEM and PTSD/CPTSD (k = 132, r = .27, 95% CI [.23, .30]) 95% PI [−.04, .53] with substantial heterogeneity (I² = 86.5%). Higher estimates were identified for emotional abuse (k = 71, r = .29, 95% CI [.25, .33]) compared to emotional neglect (k = 55, r = .24, 95% CI [.19, .28]). Effect sizes were similar for CPTSD (k = 29, r = .29, 95% CI [.24, .35]) and PTSD (k = 103, r = .26, 95% CI [.22, .30]). No differences were observed by sex, age, or population type (clinical vs. non-clinical). However, higher estimates were identified in high-income settings and more developed contexts. Narrative synthesis identified cognitive, affective, physiological and interpersonal mechanisms as potential mediators. The certainty of the evidence was generally low. Limitations include the predominance of cross-sectional designs, reliance on self-report measures of CEM and variability across outcome measures.

Conclusions: These findings underscore the importance of CEM in trauma-related symptomatology and support the need for its systematic assessment in clinical and research settings, particularly by highlighting differences between emotional abuse and emotional neglect.

HIGHLIGHTS

• Childhood emotional maltreatment, particularly emotional abuse, is associated with trauma-related outcomes, including PTSD and complex PTSD.

• Evidence on complex PTSD remains limited, highlighting the need for further research on childhood emotional maltreatment and a broader range of trauma-related outcomes.

• Evidence on underlying mechanisms is scarce: region and socioeconomic status act as moderators, while psychological and affective processes may mediate these associations.

• Large longitudinal and cross-cultural studies using harmonised assessment approaches, including both subjective and objective measures, are needed to clarify causal pathways linking childhood emotional maltreatment to PTSD and complex PTSD.

1. Introduction

Childhood emotional maltreatment (CEM) is increasingly recognised as an important risk factor for trauma-related outcomes (Gama et al., 2021; Hashim, Alimoradi, et al., 2024; Peter et al., 2026), yet its specific association with posttraumatic stress disorder (PTSD) and complex PTSD (CPTSD) remains insufficiently characterised.

PTSD affects millions worldwide and is associated with substantial functional impairment and chronic disability (Koenen et al., 2017; Ressler et al., 2026). Epidemiological data indicate a lifetime prevalence of approximately 4-6%, with rates reaching 25–30% following severe psychological trauma (Ressler et al., 2026). Similarly, pooled prevalence estimates across trauma-exposed populations approach 24% (Schincariol et al., 2024). Despite the availability of evidence-based pharmacological and psychotherapeutic treatments (Billings & Nicholls, 2025; Fares-Otero et al., 2026; Hoppen et al., 2024), many individuals experience only partial recovery, with persistent symptoms, reduced quality of life, and elevated risk of comorbidity, suicide and medical condition, including cardiovascular and metabolic diseases (Ressler et al., 2026; Sareen et al., 2007; Shea et al., 2010). These limitations highlight the need to better understand the factors associated with the onset and persistence of trauma-related disorders, to guide prevention and intervention strategies.

PTSD and CPTSD represent two key trauma-related outcomes, with CPTSD recently introduced in the International Classification of Diseases, 11th Edition (Brewin et al., 2017; Fung et al., 2020). PTSD captures core responses to traumatic exposure, characterised by re-experiencing, avoidance, hyperarousal and negative alterations in cognition and mood (McHugh & Treisman, 2007; Shalev et al., 2017). CPTSD includes additional disturbances in self-organisation, including affective dysregulation, negative self-concept and relational difficulties, occurring alongside the core symptoms of PTSD (Maercker et al., 2022). While single-event traumas such as natural disasters or accidents can precipitate PTSD, CPTSD more often arises from repeated or prolonged interpersonal trauma, including childhood abuse or violence in adulthood (Huynh et al., 2025; Karatzias et al., 2017).

CEM refers to caregiver behaviours that undermine a child´s emotional security and sense of self-worth and encompasses two main forms: emotional abuse (EA), defined as intentional acts like verbal assaults, humiliation, rejection, or terrorising and emotional neglect (EN), defined as the passive failure to meet a child’s emotional needs for attention, validation and affection (Glaser, 2002).

CEM has historically been more difficult to define, detect, and respond to than more visible forms of maltreatment. This is partly because emotionally harmful caregiving behaviours are often embedded within everyday parent–child interactions and may not be readily recognised as abusive or abnormal. Unlike forms of maltreatment that leave observable physical evidence, CEM is frequently expressed through patterns of emotional unavailability, rejection, hostility, or invalidation, making it more difficult to identify and address. The prevalence of CEM varies widely across geographic region and gender (Moody et al., 2018), contributing to inconsistent definitions and reporting practices (McGee & Wolfe, 1991). CEM has been associated with depressive symptoms (Hashim, Alimoradi, et al., 2024) and other forms of psychopathology, including substance use, somatic symptoms and interpersonal difficulties (Cloitre et al., 2009; Fares-Otero, Alameda, et al., 2023; Fares-Otero, De Prisco, et al., 2023; Hashim, Sheel, et al., 2025; Lüönd et al., 2025). However, its association with trauma-related outcomes remains less clearly established.

The Attachment-Based Developmental Model of Childhood Emotional Abuse provides a theoretical framework for understanding how CEM may increase vulnerability to PTSD and CPTSD (Riggs, 2010). According to this model, emotionally abusive or neglectful caregiving disrupts the development of a child’s sense of safety, emotional security, and trust in close relationships. Children who grow up in environments characterised by criticism, rejection, humiliation, emotional invalidation, or lack of emotional responsiveness may develop negative beliefs about themselves and others and perceive the world as unsafe or unsupportive. Over time, these experiences can interfere with the development of healthy emotional regulation and increase sensitivity to interpersonal stress and threat, and contribute to maladaptive coping strategies (Brahmand et al., 2016). As a result, individuals exposed to CEM may experience long-term emotional and interpersonal difficulties. These difficulties may increase vulnerability to PTSD symptoms as well as disturbances in self-organisation (DSO), including affective dysregulation, a negative self-concept, and relational difficulties, which are the defining features of CPTSD (Evren et al., 2011).

There is increasing recognition that CEM contributes to the development of both PTSD and CPTSD, yet the scope of prior quantitative reviews remains limited. Existing reviews have primarily focused on general psychopathology (Humphreys et al., 2020) in adolescents (Hashim, Alimoradi, et al., 2024), or cognitive functional outcomes rather than trauma-related outcomes specifically (Fares-Otero, Alameda, et al., 2023; Fares-Otero, De Prisco, et al., 2023; Fares-Otero, Pérez-Ramos et al., 2025 Acta Psychiat Scand). In addition, previous meta-analyses have largely examined broader childhood maltreatment constructs or specific forms such as sexual abuse (Dworkin et al., 2023; Ehring et al., 2014; Hashim, Iqbal, et al., 2025), with less attention to CEM as a distinct exposure. Reported associations between CEM and PTSD/CPTSD vary widely across studies, from negligible (Bateman et al., 2024) to strong effects (Cecil et al., 2017), underscoring the need for quantitative synthesis.

Moreover, the emergence of CPTSD as a distinct diagnosis in the ICD-11 highlights the need to better understand how specific forms of childhood maltreatment contribute to trauma-related symptom profiles (Brewin et al., 2017; Cloitre et al., 2009), yet the role of CEM in CPTSD remains largely unaddressed. Previous research has often focused on EA while overlooking EN (Carvalho Silva et al., 2024), despite increasing recognition of its clinical relevance. Consequently, a meta-analytic synthesis examining CEM, across its subtypes, in relation to both PTSD and CPTSD is needed.

At present, potential moderating (e.g. age, sex/gender (Fares-Otero et al., 2024), race/ethnicity, diagnosis type) or mediating (e.g. personality, biological stress, cognitive functions) factors have seldom been systematically examined. Notably, only one systematic review (Omidbakhsh et al., 2025) specifically focused on CPTSD has examined psychological mediators, identifying five broad processes (emotional processing, intrapersonal and interpersonal factors, dissociation and social factors) as potential mechanisms linking overall childhood maltreatment to CPTSD, although the evidence remains limited and based on a small number of studies.

Taken together, the literature on the relationship between CEM and PTSD/CPTSD is limited and inconsistent and no meta-analytic study to date has simultaneously examined EA and EN in relation to both PTSD and CPTSD, while exploring contextual moderators and mediators.

This meta-analysis aimed to address these gaps by synthesising the association between CEM and trauma-related outcomes. Specifically, the study quantified this association while differentiating between EA and EN examining PTSD and CPTSD separately and exploring potential sources of heterogeneity. Findings are intended to inform clinical assessment and support the development of more targeted trauma-focused interventions.

2. Methods

2.1. Protocol

The protocol was registered on PROSPERO (CRD42024598358). We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 (see Appendices SA2 and SA3) and the Meta-analysis of Observational Studies in Epidemiology (MOOSE) (see Appendix SA4) guidelines (Page et al., 2021; Stroup, 2000). Amendments to the protocol are presented in Appendix SA5. For a glossary of terms used in this work, see Appendix SA1.

2.2. Search strategy

A systematic literature search was conducted using multiple terms related to childhood maltreatment and trauma, focusing on two components: exposure (CEM) and outcome (PTSD and CPTSD) (see Appendix SA6) and drawing on previous work in the field (Hashim, Alimoradi, et al., 2024; Xiao et al., 2023). Relevant Medical Subject Headings (MeSH) and keywords were compiled. Search terms were applied to titles and abstracts and MeSH terms were included where available. We applied the OR operator within CEM and PTSD/CPTSD terms and the AND operator between exposure and outcome terms. The search was initially conducted on 30/10/2024 and updated on 17/03/2025 across the following databases: MEDLINE via PubMed, Scopus, Web of Science (Core Collection), PsycINFO and Embase via EBSCO, and ProQuest for grey literature. When updating the search to 19/03/2026, the PILOTS database was also searched, and 27 additional eligible articles were identified.

A snowballing approach (Wohlin et al., 2022) was also applied to identify additional studies meeting the inclusion criteria in the reference list of relevant studies. We performed backward and forward referencing, screened key journals and contacted active researchers for published and unpublished work.

2.3. Eligibility criteria

Peer-reviewed studies and grey literature published in English examining the association between CEM and trauma-related outcomes were considered for inclusion. No restrictions on age or publication date were applied. According to the PECO-S framework (Morgan et al., 2018), studies were included if they met the following criteria: (i) Population: Studies including clinical and community samples. Clinical samples may include individuals with PTSD or CPTSD as well as other psychiatric disorders. Studies were eligible if they assessed (i) PTSD or CPTSD diagnoses according to validated DSM (American Psychiatric Association, 2013) or ICD (World Health Organization, 1993) criteria (see diagnostic codes in Appendix SA7), or (ii) continuous measures of trauma-related symptoms in individuals with PTSD/CPTSD or in the general population; (ii) Exposure: Studies were included if they assessed CEM, including both overall CEM and specific subtypes (EA and/or EN) occurring before the age of 18 years, and assessed at the individual level. Appendix SA8 provides detailed definitions of all exposure variables; (iii) Comparison: Eligible studies included measures of CEM and compared outcomes across (i) other forms of childhood maltreatment, (ii) other types of adversity such as community violence, accidents, bereavement, or other stressful life events, (iii) adversity occurring in adulthood, or (iv) the absence of maltreatment or adversity; (iv) Outcome: PTSD and CPTSD outcomes assessed using validated instruments consistent with DSM (American Psychiatric Association, 2013) or ICD-10/11 criteria (Fung et al., 2020; World Health Organization, 1993), including (a) continuous symptom severity scores, (b) overall severity measures and (c) dichotomous diagnostic classifications (presence vs. absence of disorder); (v) Study-Design: observational studies, including cross-sectional and cohort studies of the association between CEM and PTSD/CPTSD disorder/symptoms in the general population and case–control studies comparing CEM prevalence between patients and healthy controls, or trauma-related outcomes between CEM-exposed and non-exposed individuals.

When multiple publications were retrieved for the same sample, the article reporting the largest sample size or providing the most relevant data was included. Other publications were only included when reporting data relevant to our secondary analyses (e.g. sex, age, socio-cultural differences, moderator/mediators).

Studies were excluded if they were reviews, clinical case reports, abstracts without full text, commentaries, letters to the editor without original data, conference proceedings, study protocols, or preprints.

2.4. Study selection and data extraction

Citations were managed using Zotero and Rayyan QCRI (https://rayyan.qcri.org/), which facilitated duplicate removal and screening (Ouzzani et al., 2016). Interrater agreement was assessed using the intraclass correlation coefficient (ICC) for continuous variables and Cohen’s Kappa (κ) for categorical variables. Title and abstract screening were conducted independently by three reviewers (MH, BR, AA), achieving strong inter-rater reliability (ICC = .99; κ = 0.94). After excluding non-relevant records, full-text eligibility was assessed by the same three reviewers, with at least two reviewers assessing each study. Discrepancies were resolved through discussion and consensus. In the updated search, title/abstract and full-text screening was conducted independently in duplicate: NEF-O performed one review, while the second review was conducted by four trained master’s students, achieving 95% agreement (see Acknowledgments).

Between December 2024 and May 2025, three authors (MH, BR, AA) extracted data using a predefined data extraction sheet, capturing author, year, country/region, mean age, sample size, sex and ethnicity distribution, design, sample type (clinical vs. non-clinical), CEM type, timing, duration and perpetrator, and assessment tools for CEM, PTSD and CPTSD (if reported). Continuous (symptom severity) and categorical (diagnosis) effect sizes were extracted. We extracted correlations (r) or other statistics to compute r as the common effect size. Following the search update, between March and April 2026, NEF-O extracted the new data using the same extraction sheet and a second check was performed by MH. Any disagreements were resolved through consensus. Overall, authors of 24 studies were contacted to request published, unpublished, or missing data. Nine authors provided additional information, while 15 did not respond.

2.5. Study quality and certainty of the evidence assessment

Included studies were assessed for methodological quality using a modified version of the Newcastle–Ottawa Scale (NOS) (Herzog et al., 2013). This adaptation evaluated key domains such as sample selection, group comparability and outcome or exposure measurement. Additional criteria, including sample size, control for confounders and the appropriateness of statistical analyses, were incorporated in line with the Cochrane Handbook guidelines (Higgins et al., 2011) and previous meta-analytic studies in the field (Fares-Otero, Alameda, et al., 2023; Fares-Otero, Carranza-Neira, et al., 2025; Fares-Otero, De Prisco, et al., 2023; Hashim, Osmany, et al., 2024). Quality ratings were independently conducted by three authors (MH, BR, AA) for the initial search and by two authors (MH, NEF-O) for the updated search (Appendix SA9.A). Studies scoring fewer than five points were classified as low quality; those scoring five points were considered fair quality; those scoring six points were considered good quality; and those scoring more than six points were considered high quality. Any disagreements were resolved through discussion and consensus.

Certainty in the evidence for each outcome was assessed using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) system (Brignardello-Petersen & Guyatt, 2025; Prasad, 2024), following recommendations for observational evidence (Yousefifard & Shafiee, 2023). For association questions, observational studies were considered the appropriate study design and certainty ratings therefore started at high certainty. Certainty was evaluated across the domains of risk of bias, inconsistency, indirectness, imprecision, and publication bias, together with other considerations that may increase the certainty of evidence (Appendix SA9.B).

2.6. Statistical analyses

To quantify the association between CEM and PTSD/CPTSD, Pearson correlation coefficients (r) were used as the common effect size. When not reported, other statistics were converted using recommended formulas (Rosenthal, 1994; Wilson & Lipsey, 2001), including transformations from odds ratios or regression coefficients to r. For two studies (Liu et al., 2026; Özdemir & Yazici, 2025), effect sizes were derived by computing the correlation between CEM variables and a composite PTSD/CPTSD score constructed by aggregating available outcome subscales into a single composite score. Pearson’s correlation coefficients were transformed into Fisher’s z-values to account for non-normality and to compute sampling variances using the escalc function in R (Viechtbauer, 2010). After the analyses, Fisher’s z-values were transformed back into Pearson’s r correlation coefficients to facilitate interpretation. When studies reported both continuous and binary operationalisations of CEM, continuous estimates were prioritised (Anzures-Cabrera et al., 2011; Ben-Shachar et al., 2020).

A multilevel meta-analytic approach was used to account for statistical dependency among effect sizes (Assink & Wibbelink, 2024; Pastor & Lazowski, 2018; Van den Noortgate et al., 2015). Random-effects models were used throughout, assuming that the included studies represented a random sample from a broader population of studies (Field & Gillett, 2010). Variance components were estimated using Restricted Maximum Likelihood (REML).

The multilevel model accounted for three sources of variance: (1) sampling variance (Level 1), (2) within-study variance (Level 2), and (3) between-study variance (Level 3). Model fit was evaluated using the log-likelihood, Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC), with lower values indicating better fit.

Effect sizes were nested within studies using a random intercept model to account for dependency among multiple estimates from the same sample. Confidence intervals (CIs) and prediction intervals (PIs) were calculated to assess variability and account for between-study heterogeneity (IntHout et al., 2014). Heterogeneity was quantified using τ² (tau-squared), Cochran’s Q statistic and I² partitioned into within- and between-study components in the multilevel model.

If the number of available effect sizes did not allow for a quantitative synthesis, study findings were summarised and appraised qualitatively in a narrative synthesis (Popay et al., 2005). In addition, moderators and mediators of the association between CEM and PTSD/CPTSD were synthesised narratively.

To assess the robustness of our findings and the influence of individual studies, we conducted sensitivity analyses by (1) excluding studies rated as poor quality (NOS < 5), and (2) performing leave-one-out analyses to evaluate the impact of individual studies on the pooled estimates. An additional sensitivity analysis was conducted excluding one highly influential outlier study with an extremely large sample size and a very low PTSD event rate (Abuhasira et al., 2026).

When within- or between-study variance was observed, meta-regressions were conducted to examine whether pooled estimates were influenced by continuous moderators, including year of publication, sample size, mean age, percentage of females, percentage of white participants and study quality. Categorical variables were examined using subgroup analyses, including type of CEM (EA vs. EN), diagnostic category (PTSD vs. CPTSD), population type (clinical vs. non-clinical), geographical region (Europe vs. Africa vs. the Americas), country income level (high vs. lower-middle vs. upper-middle-income countries), development status (developed vs. developing) according to WHO guidelines (Akter & Deeba, 2022), CEM measurement (CTQ vs. other measures) and study design (cross-sectional vs. non-cross-sectional) (see Appendix SA14).

Small-study effects, which may indicate publication bias, were assessed through visual inspection of contour-enhanced funnel plots and Egger’s test (Egger et al., 1997) when at least 10 studies were available. Publication bias analyses requiring study-level independence were conducted on aggregated effect sizes at the study level to account for dependency. The trim-and-fill method (Duval & Tweedie, 2000) was applied as a supplementary analysis using these aggregated study-level effect sizes to estimate the number of potentially missing studies and examine the impact on the pooled effect size.

All quantitative analyses were conducted in R version 4.5.3 (R Core Team, 2025) using the ‘metafor’ package (Viechtbauer, 2010). Multilevel meta-analyses were conducted following established methodological recommendations for dependent effect sizes (Assink & Wibbelink, 2024; Harrer et al., 2021). Cohen’s guidelines were used to interpret effect sizes, with r values of approximately 0.10 considered small, 0.30 moderate and 0.50 or higher large (Cohen, 1992).

3. Results

3.1. Study characteristics and qualitative synthesis

The search identified 11,479 records, including 11,442 from databases and 37 from other sources. After removing 1,793 duplicates, a total of 9,649 records remained for title and abstract screening. Of these, 9,481 records were excluded for not meeting the inclusion criteria, leaving 168 full-text articles assessed for eligibility from databases and 35 identified through other methods. Of these, two reports were not retrieved and 141 studies were excluded after full-text assessment. Ultimately, 62 studies were included in the final review, of which 57 were included in the quantitative synthesis. The study selection process is summarised in Figure 1. A complete list of included studies is provided in Appendix SA10 and Appendix SA11 presents the excluded studies with reasons for exclusion.

Figure 1.

PRISMA Flowchart outlining the study selection process.

The 57 studies included in the meta-analysis were conducted between 2000 and 2026, with sample sizes ranging from 72 to 288,633 participants, yielding a total sample of 323,776 participants. Twenty-two studies were conducted in U.S.A., five studies were conducted in China, four studies in Israel, three studies in Turkey, three studies in the UK, three studies in the Netherlands, two studies in South Africa and in the rest of countries one study each. One study was multicentric involving samples from different countries/regions. Thirty-nine studies were conducted in developed and high-income countries. Most studies included mixed samples, with a mean proportion of white participants of 39.6%. Participant mean age ranged from 9.5 to 72.9 years, with an overall mean age of 28.34 (SD = 12.16). Eighteen studies were conducted in children and adolescents (or youth populations) (aged 9–19 years), while the remaining 39 studies examined adult populations. The female proportion across studies was 62.8%; two studies were conducted exclusively in male participants, while eight included only female participants. Twenty studies were conducted in clinical samples, 36 studies in non-clinical (community, control, student) samples and one study in mixed samples. Most studies used cross-sectional designs (n = 48), with a smaller number employed longitudinal cohort approaches (n = 8) and one study adopted a case–control design.

Most included studies (n = 56) relied on retrospective self-report measures of CEM. Most of them (n = 43) used the Childhood Trauma Questionnaire (CTQ) (including the short form and different language versions) (Bernstein et al., 2003), followed by the Adverse Childhood Experiences (ACEs) questionnaire (Oláh et al., 2023) (n = 7) and other retrospective self-report measures, including direct questions. Only one study used a prospective longitudinal assessment approach (LONGSCAN measures) (see Appendix SA12). Perpetrator characteristics or the relational context of CEM were generally not explicitly operationalised or reported, and/or CEM was examined within broader adversity frameworks in which the perpetrator context remained unclear. Most studies (n = 53) categorised CEM into EA (k = 71) and/or EN (k = 55) except for four studies that assessed overall CEM (n = 4). All studies measured CEM occurring during childhood, defined as before the age of 18 years.

Most studies (n = 48) relied on self-report measures of PTSD/CPTSD symptoms. Only a small number of studies (n = 9) employed clinician-administered interviews, diagnostic records, mixed assessment approaches, or informant-report measures (Appendix SA12).

Across the 57 studies, 43 assessed PTSD (75.4%) only and four assessed CPTSD (7.0%) only, while 10 studies (17.5%) assessed both PTSD and CPTSD. PTSD and CPTSD were most commonly assessed using the PTSD Checklist for DSM-5 (PCL-5) (Blevins et al., 2015) (n = 17), followed by the International Trauma Questionnaire (ITQ) (Cloitre et al., 2018) (n = 11) and the Clinician-Administered PTSD Scale for DSM-5 (CAPS-5) (Weathers et al., 2018) (n = 4). Other instruments were used in one or two studies each, reflecting substantial heterogeneity in measurement approaches (see Appendix SA12). Further study details are summarised in Table 1.

Table 1.

Sociodemographic and assessment characteristics of included studies.

Authors and year of publication	Country/Region	N	Mean age (years/range)	Female (%)	Type of CEM	Instrument to assess CEM	Outcome(s)	Source/criteria/Instrument to assess PTSD/ CPTSD	NOS total	Key findings
Abu Khudair et al. (2026)	Jordan/ Middle East	3,597 4,407	8 to 11 12 to 18	57.1	EA, EN	ACE (adapted) + LEC-5	PTSD	CRIES-13	7	A high prevalence of PTSD symptoms was observed among children and adolescents, with higher rates in females and adolescents. Feeling unloved by family was associated with higher PTSD symptoms, and exposure to multiple ACEs showed a dose-response relationship, with greater numbers of ACEs increasing the likelihood of PTSD.
Abuhasira et al. (2026)	Israel/ Middle East	288,633	18	39.5	EA, EN	Modified ACE (9 items)	PTSD	ICD-10 (clinical records)	6	Sexual abuse and EN assessed before military service were associated with increased risk of combat-related PTSD, while physical abuse showed no effect.
Al Barathie et al. (2026)	Lebanon/ Middle East	426	36.9	70.7	CEM	Self-report of childhood adversities (3 items)	PTSD	PCL-5	7	CEM predicted ASD at 21–27 days and both full-threshold and subthreshold PTSD at 6–7 months.
Auslander et al. (2018)	USA/ North America	234	15	100	EA	CTQ	PTSD	CPSS	6	EA, physical and sexual abuse, as well as physical neglect, were associated with higher levels of revictimisation and increased PTSD and depressive symptoms. The association between EA and revictimisation was mediated by PTSD and depressive symptoms, with the same pattern observed for sexual abuse. The relationship between physical abuse and revictimisation was fully mediated by PTSD, while no mediation effects were found for physical neglect.
Balsam et al. (2010)	USA/ North America	669	36	62	EA	CTQ	PTSD	PTSD Checklist-civilian version	8	EA, physical, and sexual abuse were all positively associated with poorer mental health outcomes, including PTSD, anxiety, depression, and perceived stress. EA was associated with depression, perceived stress, anxiety and PTSD symptoms. The strength of associations between abuse and mental health outcomes varied by race/ethnicity, with EA more strongly related to anxiety and PTSD among African American, and physical abuse more strongly related to these outcomes among Latina/o participants.
Basting et al. (2024)	USA/ North America	721	40.8	26	EA, EN	ACE	PTSD	15 items- subscale of the PDSQ	4	Four latent classes of ACEs were identified (high ACEs, maltreatment, household problems, low ACEs). Individuals in the high ACEs class showed the highest levels of PTSD symptoms compared to all other classes, while those in the maltreatment and household problems classes also reported higher PTSD symptoms than the low ACEs class.
Basting et al. (2025)*	USA/ North America	707	40.8	26.4	EA	ACE	PTSD	PDSQ	6	PTSD symptoms moderated the association between childhood sexual abuse and CSB; sexual abuse was associated with CSB only at high levels of PTSD symptoms, and EA was associated with CSB.
Bateman et al. (2024)	UK/ Europe	72	28	89	EA, EN	CTQ	CPTSD, PTSD	PCL-5	5	Patients with co-occurring borderline personality disorder and PTSD symptoms showed higher severity of borderline symptoms, greater dissociation, and lower well-being compared to those with borderline personality disorder. Mentalising deficits were more pronounced in the comorbid group and partially mediated the associations between borderline symptoms and PTSD symptom status but did not mediate the association between sexual abuse and PTSD symptoms.
Cecil et al. (2017)	UK/ Europe	204	19	53	EA, EN	CTQ	PTSD	TSCC-A	7	CM types were highly interrelated and frequently co-occurred, and a higher number of maltreatment types was associated with greater severity of psychiatric symptoms, including internalising difficulties, PTSD symptoms, anger, and dissociation. EA was associated with internalising difficulties and trauma-related symptoms (PTSD symptoms, anger, and dissociation), whereas no associations were found for other maltreatment types.
Chesin et al. (2024)	USA/ North America	327	20	100	CEM	CTQ	PTSD	PCL-5	5	Three distinct victimisation classes were identified (low, childhood, and lifetime victimisation). PTSD symptom severity was higher in the childhood victimisation class compared to both the low and lifetime victimisation classes, indicating stronger association with CM than with combined lifetime victimisation.
Ciringione et al. (2025)	Italy/ Europe	83	≥18	91.6	EA, EN	CTQ	PTSD	PDS-3	3	EA predicted borderline symptoms; sexual abuse predicted dissociation; emotion dysregulation was a central transdiagnostic factor linking ACEs with borderline and dissociative symptoms.
Draczynska et al. (2025)	Poland/ Europe	794	30.3	91.3	EA, EN	TEC	PTSD, CPTSD	ITQ (ICD-11 for PTSD/CPTSD) PCL-5 for PTSD	5	EN was the strongest predictor of CPTSD and predicted PTSD symptom severity, while threat-related trauma (threat to life, pain, bizarre punishment) was the strongest predictor of PTSD diagnosis.
Doehrmann et al. (2025)	USA/ North America	129	18	70	EA	UCLA PTSD Reaction Index	PTSD	PCL-5	6	Sexual violence was associated with higher intolerance of uncertainty, which mediated the relationship between sexual violence and PTSD symptoms. EA was associated with PTSD symptoms at the bivariate level but did not remain a predictor in multivariable models.
ElBarazi et al. (2025)	Syria/ Middle East	508	22.6	66.3	EA, EN	CTQ	PTSD	PCL-5	4	Physical abuse, EA and sexual abuse increased the likelihood of PTSD in adulthood, while EN and physical neglect were not associated with PTSD.
Evren et al. (2011)	Turkey/ Europe-Asia	156	42	0	EA, EN	CTQ	PTSD	CAPS-5	6	Lifetime PTSD in alcohol-dependent men was associated with poorer QoL across multiple physical and mental domains. Dissociation and lifetime PTSD predicted impairment in QoL, and dissociative patients reported higher levels of CM. Among individuals with lifetime PTSD, EA was linked to poorer QoL through its association with dissociation.
Evren et al. (2016)	Turkey/Europe-Asia	190	42	0	EA, EN	CTQ	PTSD	PCL-C	6	EA, but not EN, was associated with PTSD symptoms, and both inattentive and hyperactive/impulsive ADHD dimensions were linked to PTSD severity. EA may play a mediating role in the relationship between ADHD symptoms and PTSD symptoms among male inpatients with AUD.
Faizan et al. (2025)	India/Asia	410	20.6	74.6	EA, EN	CTQ-SF	PTSD	PCL-5	8	EA and physical abuse showed the strongest associations with PTSD, depression, and anxiety, while EN predicted depressive symptoms.
Ferrajão et al. (2024)	Uganda/ Africa	401	16	51	EA	DQ	PTSD	HTQ	6	Exposure to ACEs showed associations with PTSD and CPTSD. Exposure to 2–3 and 4–5 ACEs was associated with higher likelihood of PTSD, while specific ACEs such as physical violence, witnessing injury or death, and attempted suicide were linked to PTSD. CPTSD was associated with exposure to sexual abuse, bullying and threats of violence, and with higher cumulative exposure. Additionally, fearful attachment was associated with PTSD, while dismissing attachment was associated with CPTSD.
Fitzgerald and Lunt (2024)	USA/ North America	219	29.2	100	EA, EN	CTQ	PTSD	PTSD checklist civilian version	5	Facets of dispositional mindfulness, particularly nonjudgment and nonreactivity, mediated the association between EA/nEN and PTSD symptoms, with nonjudgment showing the strongest effect.
Florez et al. (2022)	USA/ North America	172	34	100	EA	CTQ	PTSD	DTS	3	Childhood physical, EA, and sexual abuse were not associated with alcohol misuse. PTSD severity and hyperarousal mediated the associations between all types of childhood abuse and alcohol misuse, while avoidance mediated the associations for physical and EA but not for sexual abuse. Re-experiencing symptoms did not mediate any associations.
Haim-Nachum, Amsalem, et al. (2024a)	USA/ North America	230	46	52	EA, EN	CTQ	PTSD	PCL-5	7	CM was associated with anhedonia, which in turn was associated with higher severity of PTSD and depressive symptoms but not social anxiety. Anhedonia mediated the relationship between CM and both PTSD and depressive symptoms. At the subtype level, this mediation was observed for EA and sexual abuse, as well as for EN and physical neglect, but not for physical abuse.
Haim-Nachum, Lazarov, et al. (2024b)	USA/ North America	642	37	63	EA, EN	DQ	PTSD	PC-PTSD-5	5	CEM was associated with depression and anxiety but not PTSD, and physical maltreatment was associated only with PTSD. The dimensional classification approach showed better performance in predicting symptoms compared to the abuse/neglect approach.
Haim-Nachum et al. (2024)*	USA/ North America	642	37	63	EA, EN	DQ	PTSD	PC-PTSD-5	5	Self-stigma mediated the relationship between CM and symptoms of PTSD, depression, and anxiety. This mediation was observed for both abuse and neglect. At the subtype level, mediation effects were present for sexual abuse and for both EN and physical neglect, but not for physical abuse or EA.
Harb et al. (2025)	USA/ North America	160	33.7	56.3	EA, EN	CTQ	PTSD	CAPS-5	7	Sleep disturbances mediated the relationship between EA, EN, and physical neglect and later PTSD symptom severity after adult trauma.
Hilberdink et al. (2023)	The Netherlands / Europe	568	73	52	EA, EN	CTQ	PTSD	LEC-5	6	CM was associated with probability of classification into more severe psychological symptom profiles, including profiles characterised by PTSD symptoms, in both men and women. EA was associated with increased likelihood of mild symptom profiles in both sexes, while EN was not associated with profile assignment.
Ho & Su (2026)	Taiwan/ Asia	669	35.7	53.5	EA, EN	CTQ	PTSD, CPTSD	ITQ	8	CM was strongly associated with PTSD and CPTSD symptoms, with indirect effects observed through insecure attachment and emotion dysregulation.
Kim & Ko (2025)	South Korea/ Asia	249	28.8	78.7	EA, EN	CTQ	CPTSD	ITQ	4	EN and EA showed the strongest connections with CPTSD symptoms. Network analysis suggested a pathway from EN to EA, with negative self-concept bridging trauma experiences and CPTSD symptoms.
Kühner et al. (2025)	The Netherlands/ Europe	197	38	81.2	EA, EN	CTQ-SF	PTSD	CAPS-5	6	Sexual abuse and EA were associated with PTSD severity in multivariable analyses, whereas EN and other maltreatment types were not.
Lahav et al. (2026)	Israel/ Middle East	404	36.7	100	EA	CTQ-SF	PTSD, CPTSD	ITQ	5	EA was associated with higher levels of PTSD, DSO, and BDD symptoms and emerged as a transdiagnostic risk factor, whereas sexual abuse was associated with PTSD symptoms only and physical abuse showed no effect.
Lawrence et al. (2025)	USA/ North America	596	12–18	49.3	EA	LONGSCAN measure	PTSD	TSC	7	EA, physical abuse, and sexual abuse were associated with PTSD symptoms at age 12 and 16, with indirect effects observed via PTSD at age 12 and deviant peer affiliation at age 14, as well as via adolescent abuse.
Liu et al. (2026)	China/ Asia	2,063	21.3	41.5	EA, EN	ACE	PTSD, CPTSD (DSO)	ITQ	7	Three distinct joint trajectories of PTSD and DSO symptoms were identified: low symptom, increasing DSO with rebounding PTSD, and moderate DSO with decreasing PTSD. EN consistently predicted higher-risk trajectories, while physical abuse and parental incarceration were linked to the most severe trajectory pattern.
Lortye et al. (2024)	The Netherlands/Europe	209	37	46	EA, EN	CTQ	PTSD, CPTSD	CAPS-5 for PTSD, ITQ for CPTSD	8	Overall CM and its types were associated with PTSD severity in univariate analyses, EA and SA were associated with PTSD severity. EA was associated with CPTSD classification and severity.
Marques-Feixa et al. (2023)	Spain/ Europe	187	13	57	EA, EN	CTQ	CPTSD	CBCL	7	CM was associated with increased internalising, externalising, and other psychopathological symptoms, poorer global functioning, and greater psychiatric comorbidity. CM was also associated with higher levels of all CPTSD domains, including PTSD symptoms, emotion dysregulation, negative self-concept, and interpersonal problems. All CM subtypes were associated with CPTSD symptomatology, with EA and EN showing associations across domains. The timing of exposure showed differential effects, with specific developmental periods moderating the associations between CM subtypes and CPTSD domains.
McCutchen et al. (2022)	USA/ North America	2,859	44	52	EA, EN	ACE	PTSD, CPTSD	ITQ	7	CM was associated with increased likelihood of meeting diagnostic criteria for PTSD and CPTSD. EN was the only subtype associated with PTSD when controlling for other ACEs, while CPTSD was associated with multiple ACEs, including sexual abuse, EN, and household dysfunction. Increasing number of ACEs was associated with higher likelihood of PTSD and particularly CPTSD, with effects observed for four or more ACEs.
Mirhashem et al. (2017)	UK/ Europe	84	35	46	EA, EN	CTQ	PTSD, CPTSD	PCL-5	5	EA was associated with higher PTSD symptoms. PTSD symptoms were associated with substance-related problems, and EA showed an indirect association with substance-related problems through PTSD.
Mutiso et al. (2025)	Kenya/ Africa	1,972	20.5	55	EA, EN	TADS	PTSD	HTQ	7	ACEs were positively associated with PTSD, depression, and suicidality. Resilience both moderated and mediated these relationships, reducing the impact of ACEs, particularly for PTSD and depression, with more limited effects on suicidality.
Moore et al. (2013)	Australia/ Oceania	291	17	12	CEM	CTQ	PTSD	K-SADS-PL	8	CM was highly prevalent among young offenders and was associated with increased likelihood of PTSD diagnosis. All CM subtypes, including EA and EN, were associated with PTSD in unadjusted analyses. Experiencing multiple forms of CM showed association, with exposure to three or more severe CM subtypes markedly increasing the likelihood of PTSD.
Nagar & Nakash (2024)	Israel/ Middle East	170	37	59	EA	CTQ	PTSD	SCID-I/P, SOCD-II	5	CM was associated with adult psychopathology, with associations observed between emotional, physical, and sexual abuse and clinician-rated borderline personality disorder. PTSD diagnosis was associated with sexual abuse as reported by patients and interviewers, while no associations were found between CM and mood or anxiety disorders. EA and physical abuse were associated with poorer adaptive functioning.
Negriff (2020)	USA/ North America	454	18	33	EA, EN	CTI	PTSD	YSSC	5	CM was associated with higher levels of depressive, anxiety, PTSD symptoms, and externalising behaviour, with maltreatment showing stronger associations with all outcomes than household dysfunction. EA and EN were associated with all mental health outcomes, including PTSD symptoms.
Ni et al., (2022)	China/ Asia	5,195	14	49	EA, EN	CTQ	PTSD	CRIES	8	EA and negative life events were associated with higher PTSD symptoms. A dose-response pattern was observed, with greater exposure to multiple stressors associated with higher PTSD and depressive symptoms.
Nöthling et al. (2019)	South Africa/ Africa	215	15	61	EA, EN	CTQ	PTSD	CPC	3	EA, sexual abuse, physical abuse, physical neglect, community violence, and trauma load were positively correlated with PTSD symptom severity, whereas EN was not associated with PTSD.
Olgiati & Serretti (2022)	USA/ North America	482	43	68	EA	DQ	PTSD	Clinical record	4	CEM, parental neglect, and physical abuse were more frequently reported in individuals with PTSD compared to those without PTSD. PTSD was associated with higher depression severity, work and social impairment, suicidality, and comorbid anxiety disorders. EA was associated with higher bipolar spectrum score and was associated with mood activation during antidepressant treatment.
Özdemir and Yazici (2025)	Turkey/Europe-Asia	366	30.2	85	EA, EN	CTQ-SF	PTSD	PCL-5	4	CM was positively associated with PTSD symptoms, and this relationship was mediated by attachment anxiety but not attachment avoidance.
Porat-Butman et al. (2025)	Israel/ Middle East	114	28.3	69.5	EA, EN	CTQ	PTSD	PCL-5	4	CM was associated with impaired belief updating about strangers, and poorer updating strengthened the association between maltreatment and PTSD symptom severity.
Price et al. (2017)	USA/ North America	82	35	46	EA, EN	CTQ	PTSD	PCL-5	7	CM was associated with higher PTSD symptom severity. EA, EN, and sexual abuse moderated the association between positive urgency and PTSD symptoms, such that the relationship between positive urgency and PTSD was stronger at lower levels of maltreatment and diminished as maltreatment severity increased.
Rosenek et al. (2025)	International / Multi-regions	159	37.4	73.6	EA, EN	CTQ	PTSD	PCL-5	7	EA and physical abuse were associated with pain intensity and PTSD symptoms. PTSD symptoms (re-experiencing, hyperarousal) were also linked to higher pain. Chronic pain individuals showed stronger associations between trauma, PTSD, and coping strategies, with more complex trauma-coping dynamics.
Santelices et al. (2024)	Chile/ South America	2,101	30	57	EA, EN	ACE	PTSD	DQ	6	EN showed a weak association with PTSD. Cumulative ACE exposure showed a dose-response relationship, with increasing number of ACEs associated with higher likelihood of PTSD, particularly for two or more ACEs.
Schiff et al. (2023)	USA/ North America	249	15	100	EA, EN	CTQ	PTSD	CPSS	6	EA, physical abuse, and sexual abuse were associated with higher PTSD symptoms, while physical neglect was not associated with PTSD symptoms. CM, including EA, was also associated with higher depressive symptoms. PTSD and depression mediated the associations between abuse (including EA) and poorer self-rated physical health, whereas no mediation effects were observed for physical neglect.
Sheinbaum et al. (2026)	Mexico	1,612	30.3	79.4	EA, EN	CTQ-SF	PTSD	PCL-5	4	EA was associated with all outcomes except negative schizotypy, EN was associated with depression, anxiety, and schizotypy, and sexual abuse was associated with PTSD symptoms.
Sullivan et al. (2006)	USA/ North America	89	16	65	EA, EN	CTQ	PTSD	CPSS	4	EA and sexual abuse were associated with higher PTSD symptoms. EA was associated with PTSD symptom clusters (reexperiencing, avoidance/numbing, and arousal) and overall PTSD severity. Sexual abuse was associated with reexperiencing symptoms only. Physical abuse, EN, and physical neglect were not associated with PTSD symptoms when all CM subtypes were considered simultaneously.
Tabb et al. (2022)	USA/ North America	147	50	33	EA, EN	ACE	PTSD	PCL-5	4	ACEs were associated with higher PTSD symptoms, with both individual and cumulative ACE exposure showing positive associations with PTSD severity and provisional PTSD. EA and EN were associated with increased PTSD symptoms. A dose-response relationship was observed, with higher cumulative ACE exposure associated with greater PTSD symptom severity and higher likelihood of provisional PTSD.
Theran & Han (2013)	USA/ North America	257	20	100	CEM	CTQ	PTSD	TSC–33	3	CEM was associated with higher depression, lower self-esteem, and higher traumatic symptoms. Authenticity in close relationships partially mediated the association between emotional CM and depression, self-esteem, and traumatic symptoms, while authenticity with parents and non-family members did not mediate these associations.
Thompson et al. (2000)	USA/ North America	335	32	100	EA, EN	CTQ	PTSD	NWS-PTSD module	4	CM was associated with higher likelihood of current PTSD, with women in the suicide attempt group more likely to report both CM and PTSD than controls. EA and EN were both associated with PTSD.
Tian et al. (2022)+	China/ Asia	478	17	55	EA, EN	CTQ	CPTSD	ITQ	5	CM was associated with distinct symptom profiles, including PTSD and CPTSD. EA was associated with increased likelihood of belonging to the CPTSD class compared to the PTSD class, while physical abuse, physical neglect, EN, and sexual abuse were not associated with class membership. The CPTSD class showed higher levels of depression, anxiety, and stress, and lower levels of life satisfaction and physical health compared to other classes.
Tian et al. (2025)+	China/ Asia	395	17.5	54.2	EA, EN	CTQ	PTSD, CPTSD (DSO)	ITQ	7	EA predicted PTSD and DSO symptoms via attachment anxiety, while emotional and physical neglect predicted DSO symptoms via attachment avoidance.
van den Heuvel et al. (2023)	South Africa/ Africa	262	15	63	EA, EN	CTQ	PTSD	CPC	6	Sexual abuse was associated with internalising disorders, but this association was no longer present when PTSD severity was included. Physical abuse was associated with externalising disorders, and this association remained after including PTSD severity. Physical abuse, EN, and PTSD severity were associated with comorbid internalising and externalising disorders.
Watts et al. (2021)*	USA/ North America	396	24	80	CEM	CTQ	PTSD	SPRINT	7	CEM was associated with higher PTSD symptoms. CEM was also associated with centrality of the event and intrusive rumination, and both were associated with PTSD symptoms.
Watts et al. (2025a)*	USA/ North America	263	27.2	76.8	CEM	CTQ	PTSD, CPTSD	ITQ	5	CEM was associated with PTSD symptoms, with posttraumatic appraisals mediating this relationship.
Watts et al. (2025b)	USA/ North America	288	27	76.7	CEM, EA, EN	CTQ	PTSD, CPTSD	ITQ	5	CEM showed the strongest association with PTSD and DSO symptom severity and was the only predictor of CPTSD diagnosis. EA predicted CPTSD diagnosis and PTSD symptoms, while EA and EN were associated with DSO symptoms.
Watts et al. (2025c)*	USA/ North America	387	23.7	80.1	CEM	CTQ	PTSD	SPRINT	6	CEM was the most substantial predictor of negative meta-emotions. Negative meta-emotions predicted PTSD symptoms and explained 42.6% of the variance in PTSD symptom severity among those exposed to CEM.
Xie et al. (2023)	USA/ North America	109	34	69	EA, EN	CTQ	PTSD	PCL-5	7	ACE was positively associated with PTSD symptoms at both 2 weeks and 12 months post-trauma. ACE was negatively associated with right and left hypothalamus volume at 2 weeks, and with right hypothalamus volume at 12 months. Smaller right hypothalamus volume at 2 weeks was associated with greater PTSD symptom severity at 12 months. Right hypothalamus volume at 2 weeks partially mediated the association between ACE and PTSD symptoms at 12 months.
Zhou et al. (2025)	China/ Asia	1,894	13.9	48.5	EA, EN	CTQ-SF	CPTSD	ITQ-CA	8	EA was predictor of CPTSD symptoms. Mindfulness contributed the most among positive psychological traits, followed by self-compassion, while gratitude showed no significant association with CPTSD symptoms.

Note: Abbreviations: ACE: Adverse Childhood Experience; ADHD: Attention-Deficit/Hyperactivity Disorder; ASD: Acute Stress Disorder; AUD: Alcohol Use Disorder; BDD: Body Dysmorphic Disorder; BPD: Borderline Personality Disorder; CAPC: The Child and Adolescent PTSD Checklist; CAPS-5: Clinician-Administered PTSD Scale for DSM-5; CBCL: Child Behaviour Checklist; CEM: Childhood Emotional Maltreatment; CM: Childhood maltreatment; CPC: The Child PTSD Checklist; CPSS: The Child PTSD Symptom Scale; CPTSD: Complex Post Traumatic Stress Disorder; CRIES: Children’s Revised Impact of Event Scale; CSB: Compulsive Sexual Behaviour; DQ: Direct Questions; DSO: Disturbances in Self-Organisation; DTS: Davidson Trauma Scale; EA: Childhood Emotional Abuse; EN: Childhood Emotional Neglect; HTQ: Harvard Trauma Questionnaire; IPV: Intimate Partner Violence; ITQ: The International Trauma Questionnaire; ITQ-CA: International Trauma Questionnaire – Child and Adolescent Version; K-SADS-PL: Kiddie Schedule for Affective Disorders for Children – Present and Lifetime Version; LEC-5: Life Events Checklist for DSM-5, LGB = Lesbian, Gay and Bisexual; LONSCAN: Longitudinal Studies of Child Abuse and Neglect; NWS: National Women’s Study; PC-PTSD-5: Primary Care PTSD Screen for DSM-5; PCL-5: PTSD Checklist for DSM-5; PDS-3: Posttraumatic Stress Diagnostic Scale; PDSQ: Psychiatric Diagnostic Screening Questionnaire; QoL: quality of life; PTSD: Post Traumatic Stress Disorder; QoL: Quality of Life; SCID-I/P and SOCD-II: Structured Clinical Interview for DSM-IV Axis I and Axis II Disorders; SCID: Structured Clinical Interview for DSM Disorders; SES: Socioeconomic Status; SPRINT: Short Post-Traumatic Stress Disorder Rating Interview; TADS: Trauma and Distress Scale; TSC-33: The Trauma Symptom Checklist; TSCC-A: Trauma Symptom Checklist for Children – Alternate version; YSSC: Youth Symptom Survey Checklist. Studies marked with an asterisk (*) and shaded in grey met the inclusion criteria and were included in the systematic review but were excluded from the meta-analysis due to overlapping samples. Articles marked with a cross (+) indicate studies based on overlapping samples, yet retained in the meta-analysis as they contributed non-overlapping outcome data (PTSD and CPTSD).

3.2. Qualitative review of moderators and mediators

Ten (16.1%) of the 62 included studies in review reported evidence of moderation or mediation in the pathway between CEM and PTSD and/or CPTSD. Of these, three studies (Balsam et al., 2010; Cecil et al., 2017; Lortye et al., 2024) identified moderation effects, whereas seven studies (Fitzgerald & Lunt, 2024; Haim-Nachum, Amsalem, et al., 2024; Haim-Nachum, Lazarov, et al., 2024; Harb et al., 2025; Tian et al., 2025; Watts et al., 2021, 2025) reported mediation effects (see Figure 2). A full description is provided in Appendix SA13.

Figure 2.

Moderators and mediators in the association between childhood emotional maltreatment (CEM) and trauma-related outcomes. Abbreviations: PTSD = Posttraumatic stress disorder; CPTSD = Complex posttraumatic stress disorder.

Note. Evidence for moderation/mediation effects is supported by one study each. Dashed lines indicate associations that are not based on longitudinal evidence and therefore do not imply confirmed causal direction.

3.3. Study quality and certainty of the evidence

Overall, the studies included in the review were rated as having fair methodological quality (Mean NOS rating = 5.77, SD = 1.45). According to the NOS classification, 21 studies were rated as high quality, 11 studies were rated as good quality, 12 studies as fair quality and 13 studies as low quality, mainly due to lack of representativeness, failure to report non-respondents and limited use of covariates (see Appendix SA9.A).

According to the GRADE framework, the certainty of evidence was generally low, primarily due to concerns regarding risk of bias and substantial heterogeneity across studies. Certainty for CPTSD-specific findings was rated as very low because of additional concerns regarding indirectness and uncertainty surrounding publication bias, reflecting the smaller evidence base and the use of proxy measures in some studies (see Appendix SA9.B).

3.4. Meta-analyses

The multilevel meta-analysis included 132 effect sizes from 57 studies (see Figure 3 and Figure SA1 in Appendix SA14 for the forest plot with the correlations extracted or calculated from the original articles). The pooled effect size indicated a positive association between CEM and PTSD/CPTSD (r = .27, 95% CI [.23, .30]). The 95% PIs ranged from −.04 to .53, indicating substantial variability in true effects across studies.

Figure 3.

Associations between CEM and trauma-related outcomes.

Note. The forest plot depicts the overall association between CEM and PTSD/CPTSD using a three-level meta-analytic model; r correlations and their corresponding 95% CIs in each single study are not the r correlations extracted or calculated from the original articles; rather, they were derived from a multilevel model aggregation process.

The random-effects multilevel model estimated using REML, produced a log-likelihood of 62.13, AIC = −118.26 and BIC = −109.68. Heterogeneity was observed at both the between-study (τ² = .017) and within-study (τ² = .009) levels. Cochran’s Q test indicated considerable heterogeneity across effect sizes Q (129) =   5240.85, p < .001. A large proportion of the total variability was attributable to heterogeneity (I² = 86.5%), with 57.4% at the between-study level and 29.1% at the within-study level.

3.5. Meta-regression and subgroup analyses

Meta-regression analyses examined the effects of year of publication, sample size (log-transformed), sex (% female), age, race/ethnicity and study quality (NOS rating). No effects were observed for any of the examined moderators. See SA15 for full meta-regression results.

Subgroup analyses indicated stronger associations for EA (r = .29, SE = .02, 95% CI [.25, .33]) compared to EN (r = .24, SE = .02, 95% CI [.19, .28]). Similar effect sizes were observed for CPTSD (r = .29, SE = .03, 95% CI [.24, .35]) and PTSD (r = .26, SE = .02, 95% CI [.22, .30]). Regional variation was observed, with European (vs. non-European) samples (r = .29, SE = .05, 95% CI [.20, .38]), high (vs. low/ upper-medium) income countries (r = .27, SE = .03, 95% CI [.22, .32]) and developed (vs. developing) status (r = .28, SE = .03, 95% CI [.23, .32]), showing the highest effect sizes. No clear differences were observed between clinical and non-clinical samples. Studies using the CTQ reported stronger associations (r = .30, SE = .02, 95% CI [.27, .34]) compared to those using other measures (r = .18, SE = .03, 95% CI [.11, .24]). Cross-sectional studies showed larger effects (r = .28, SE = .02, 95% CI [.24, .31]) compared to non-cross-sectional designs (r = .20, SE = .05, 95% CI [.10, .29]). Additional subgroup results are provided in SA16.

3.6. Sensitivity analysis

Excluding studies rated as poor quality (NOS < 5) did not alter the pooled effect size (k = 105, n = 44, r = .27, 95% CI [.22, .31]). Leave-one-study-out analyses indicated that no single study had a disproportionate influence on the pooled effect size, with estimates ranging from .26 to .27 across iterations. Results were also unchanged after excluding the outlier study (Abuhasira et al., 2026).

3.7. Small study effects and publication bias

Visual inspection of the contour-enhanced funnel plot suggested some asymmetry and greater dispersion among less precise studies (see Figure SA2 in Appendix SA17). However, Egger’s regression test was not associated with small-study effects (β = .92, 95% CI [−.55, 2.38]). In addition, the trim-and-fill method did not identify any missing studies (k0 = 0) and the pooled effect size remained unchanged after adjustment (r = .27, 95% CI [.23, .30]).

4. Discussion

The present meta-analysis examined the association between CEM and trauma-related outcomes, including PTSD and CPTSD outcomes. Overall, we observed a small-to-moderate positive association between CEM and trauma-related outcomes, indicating that greater exposure to CEM is consistently linked to higher levels of post-traumatic symptoms. Although the effect size was modest, it is consistent with those typically observed in psychiatric and psychological research, particularly for distal and multifactorial risk factors. Given the high prevalence of CEM, even small-to-moderate effects may have substantial public health and clinical relevance (Mathews & Dube, 2025).

Effect sizes were similar for CPTSD and PTSD. However, given the smaller number for CPTSD (k = 29) compared to PTSD (k = 103) and the inclusion of two studies assessing only DSO as a proxy (k = 7) (Lahav et al., 2026; Liu et al., 2026), these findings should be interpreted with caution. Nonetheless, including CPTSD extends previous work that has primarily focused on PTSD and allows a preliminary comparison across diagnostic frameworks.

Notably, EA showed stronger associations than EN, suggesting that subtypes of CEM may differentially contribute to trauma-related psychopathology (Florez et al., 2022; Messman-Moore & Bhuptani, 2017; Watts et al., 2021). This pattern may reflect the relatively more direct impact of emotionally abusive experiences (Dye, 2019) on self-concept, affect regulation and perceived interpersonal threat, processes that are central to the development and maintenance of trauma-related symptoms.

The observed associations varied across studies, with stronger effect sizes identified in studies conducted in Europe and in developed high-income countries. These findings may reflect broader sociocultural and methodological differences across settings (Golden, 2020; Hansford & Jobson, 2022). In high-income and more developed countries, greater awareness of emotional forms of maltreatment and broader access to psychological services may facilitate the recognition, reporting and identification of emotionally abusive or neglectful experiences. In addition, self-report measures, such as the CTQ, may capture emotional experiences differently across cultural contexts depending on norms surrounding emotional expression, parenting practices, interpersonal communication and cultural models of self and relationships (Mesquita & Walker, 2003). There is evidence that experiences of some maltreatment types, including neglect, may be associated with the perception that these are more acceptable within one´s community. In turn, the perceived acceptability of maltreatment may represent an important mechanism through which maltreatment is either prevented or perpetuated (Wadji et al., 2023). In lower-income or developing settings, CEM may be less readily labelled as abusive, or may be overshadowed by other forms of adversity, including deprivation, community violence, or physical threat (Beatriz & Salhi, 2019). Alternatively, exposure to multiple and chronic adversities may foster resilience processes that attenuate the relative impact of CEM on later trauma-related outcomes. This interpretation is consistent with conceptualisations of resilience as successful adaptation despite significant adversity and with evidence suggesting that not all individuals exposed to childhood maltreatment experience the same degree of adverse psychosocial outcomes (Fares-Otero, Carranza-Neira, et al., 2025; Fares-Otero, J, et al., 2023). Subgroup differences may also reflect variability in reporting practices, trauma conceptualisation and measurement across socio-cultural contexts. Future cross-cultural studies using harmonised assessment approaches (Fares-Otero & Seedat, 2024) are needed to clarify these findings.

Stronger associations observed in studies using the CTQ vs other measures may reflect differences in the way CEM is assessed across instruments. The CTQ is one of the most widely used and psychometrically validated measures of childhood maltreatment and includes specific subscales for EA and EN, which may increase sensitivity to detecting CEM-related effects. In contrast, broader adversity measures or less detailed assessments may underestimate emotional forms of childhood maltreatment or fail to adequately capture their severity. Similarly, stronger effects observed in cross-sectional studies may partly reflect the influence of retrospective recall, as retrospective reports of childhood maltreatment may reflect the subjective appraisal and recollection of adverse experiences and tend to show stronger associations with psychopathology than prospective measures (Baldwin et al., 2024). Nonetheless, longitudinal studies and assessment approaches (Teicher & Parigger, 2015) capturing the perpetrator relationship and broader forms of CEM, including peer emotional victimisation, as well as the timing and chronicity of exposure (Fares-Otero & Seedat, 2026), are needed to improve understanding of how different forms of CEM relate to trauma-related outcomes over time.

The lack of moderation by sex observed in our analyses is consistent with previous research showing that sex may not influence the associations between childhood maltreatment and PTSD or CPTSD (Fares-Otero et al., 2024). Similarly, we found no evidence that age moderated the association between CEM and trauma-related outcomes. Although ICD-11 PTSD has been found to occur more frequently among women and to show a general pattern of lower prevalence in older age groups, whereas evidence for sex and age differences in CPTSD remains less consistent (McGinty et al., 2021), we found no evidence that age or sex moderated the association between CEM and trauma-related outcomes. However, our analyses only examined age as a study-level moderator and did not account for potential cohort, period, or developmental timing effects. In addition, much of the evidence derived from samples of individuals in early/emerging adulthood, which may have limited the ability to detect age-related differences. Longitudinal studies are needed to better examine how the timing of CEM exposure and developmental stage may influence PTSD and CPTSD symptoms across the lifespan.

Evidence from the included studies further suggests that the association between CEM and trauma-related outcomes may be modified by race/ethnicity and partly explained by multiple mediating pathways, particularly involving affective (e.g. anhedonia), interpersonal (e.g. attachment style) and cognitive mechanisms (e.g. self-stigma, rumination), although findings were heterogeneous.

4.1. Strengths and limitations

This study has several strengths. It provides a comprehensive synthesis involving published and grey literature on the association between CEM and trauma-related outcomes, including CPTSD, distinguishes between EA and EN and applies multilevel meta-analytic techniques to account for dependency and make efficient use of all available effect sizes (Van den Noortgate et al., 2015). In addition, beyond assessing the methodological quality of individual studies, we evaluated the certainty of the evidence using the GRADE approach for observational studies, an assessment that remains relatively uncommon in meta-analyses of observational research (Yousefifard & Shafiee, 2023). However, several limitations should be considered. The predominance of cross-sectional designs limits causal inference and precludes conclusions about temporal directionality. In addition, the reliance on self-report measures may introduce recall and reporting biases, particularly in the assessment of childhood experiences. Variability in the operationalisation and measurement approaches of CEM and trauma-related outcomes may also have contributed to heterogeneity across studies, although most studies assessed the CEM exposure and trauma-related outcomes using robust instruments. The present meta-analysis did not compare the strength of associations between CEM and other forms of childhood maltreatment or other adversities. Future studies should examine whether CEM has independent, additive, or differential associations with PTSD and CPTSD when other forms of childhood and adulthood adversity are considered simultaneously. We could not differentiate between meta-analytic results based on subjective versus objective maltreatment reports and retrospective versus prospective designs, which might show diverging effect sizes. Lastly, age of reporting PTSD/CPTSD symptoms (e.g. during adolescence, early adulthood, middle or older adulthood) may also influence effect sizes. It was also impossible to account for all the possible variations across populations with different social environments, health conditions, comorbidities and diagnoses. Furthermore, the evidence base for CPTSD was more limited compared to PTSD, restricting conclusions regarding differential associations between these diagnostic constructs. Finally, the most commonly used formulae for conversion between r, d and OR based on the assumption of equal sample size have the potential to introduce bias. Also, the inclusion of English-language studies only may have led to language bias and the exclusion of relevant evidence published in other languages.

4.2. Clinical implications and future research lines

These findings highlight the importance of systematically assessing CEM in both research and clinical settings. In particular, differentiating between EA and EN, as well as considering severity and chronicity, may improve the identification of individuals at risk for trauma-related outcomes. Given that CEM is often less visible than other forms of maltreatment (Fares-Otero & Seedat, 2024; Reading et al., 2009), it may be under-recognised in clinical practice despite its meaningful association with trauma-related symptoms. These findings also suggest that early identification of EA may represent a potential target for preventive, educational and early intervention efforts (DeJong et al., 2022; Kumari, 2020), although interventional studies are needed to clarify its potential impact on trauma-related outcomes.

Future research should prioritise longitudinal and prospective designs to clarify causal pathways and temporal dynamics linking CEM to trauma-related outcomes. Further work is also needed to improve the measurement of CEM (and other adversities), including more precise assessment of subtypes, severity, developmental timing, as well as exploration of the perpetrator relationship (familial vs. other; father vs. mother) (Fares-Otero, Schäfer, et al., 2025). Additionally, research should examine the association between CEM and a broader range of trauma-related outcomes, such as dissociation, shame, or threat perception. Finally, investigating underlying mechanisms, particularly emotion regulation, self-concept, interpersonal functioning, resilience (Fares-Otero, Carranza-Neira, et al., 2025) and neurobiological pathways, is critical for clarifying how CEM contributes to trauma-related outcomes.

5. Conclusions

This multilevel meta-analysis provides evidence of a consistent small-to-moderate positive association between CEM and trauma-related outcomes, including PTSD and CPTSD. Emotional abuse showed stronger associations than emotional neglect, highlighting the differential impact of specific CEM subtypes. Associations were more pronounced in high-income and more developed sociocultural contexts compared to lower-income settings. Although the magnitude of the association was modest, its consistency across age groups, sexes and clinical and non-clinical samples underscores the clinical relevance of CEM as a transdiagnostic risk factor for trauma-related outcomes. While evidence from the included studies suggests that this relationship may be partially explained by multiple mediating pathways, these mechanisms remain heterogeneous and insufficiently understood. Overall, our findings reinforce the importance of systematically assessing CEM in both research and clinical practice and underscore the need for longitudinal, cross-cultural and mechanistic studies to clarify causal pathways and inform prevention, educational strategies and targeted interventions.

Open Scholarship

This article has earned the Center for Open Science badges for Open Data, Open Materials and Preregistered+. The data and materials are openly accessible at https://www.crd.york.ac.uk/PROSPERO/view/CRD42024598358 and https://github.com/nat-fares/meta-cem-ptsd.git.

Conflict of interest disclosures

SS has received grants from Lundbeck, Sanofi-Aventis, and Servier outside of the submitted work. The other authors report no financial relationships with commercial interests.

Declaration of AI-assisted technologies

We used ChatGPT (GPT-5.5; OpenAI, San Francisco, CA, USA) and Claude (Sonnet 4.6; Anthropic, San Francisco, CA, USA) to assist with code exploration and data organisation, as well as to revise parts of this manuscript for improving readability. The authors carefully reviewed, verified, and edited all AI-assisted outputs and take full responsibility for the final code, the content of the manuscript, and its scientific conclusions.

Data availability

MH and NEF-O have full access to all the data in the study. Open materials: https://github.com/nat-fares/meta-cem-ptsd.git

Supplemental Material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/20008066.2026.2694185.