CoChat

Childhood Maltreatment and Mental Health Problems: A Systematic Review and Meta-Analysis of Quasi-Experimental Studies

Open access
Baldwin, Jessie R., Wang, Biyao, Karwatowska, Lucy, Schoeler, Tabea, Tsaligopoulou, Anna, Munafò, Marcus R., Pingault, Jean-Baptiste · 2023 · American Journal of Psychiatry

The DOI matches its Crossref record; 84 of the 84 references it cites were found in a registry or an index; 31 of the 53 sources its claims cite were checked for support.

Abstract

OBJECTIVE: Childhood maltreatment is associated with mental health problems, but the extent to which this relationship is causal remains unclear. To strengthen causal inference, the authors conducted a systematic review and meta-analysis of quasi-experimental studies examining the relationship between childhood maltreatment and mental health problems. METHODS: A search of PubMed, PsycINFO, and Embase was conducted for peer-reviewed, English-language articles from database inception until January 1, 2022. Studies were included if they examined the association between childhood maltreatment and mental health problems using a quasi-experimental method (e.g., twin/sibling differences design, children of twins design, adoption design, fixed-effects design, random-intercept cross-lagged panel model, natural experiment, propensity score matching, or inverse probability weighting). RESULTS: Thirty-four quasi-experimental studies were identified, comprising 54,646 independent participants. Before quasi-experimental adjustment for confounding, childhood maltreatment was moderately associated with mental health problems (Cohen's d=0.56, 95% CI=0.41, 0.71). After quasi-experimental adjustment, a small association between childhood maltreatment and mental health problems remained (Cohen's d=0.31, 95% CI=0.24, 0.37). This adjusted association between childhood maltreatment and mental health was consistent across different quasi-experimental methods, and generalized across different psychiatric disorders. CONCLUSIONS: These findings are consistent with a small, causal contribution of childhood maltreatment to mental health problems. Furthermore, the findings suggest that part of the overall risk of mental health problems in individuals exposed to maltreatment is due to wider genetic and environmental risk factors. Therefore, preventing childhood maltreatment and addressing wider psychiatric risk factors in individuals exposed to maltreatment could help to prevent psychopathology.

Overview

CoChat Summary

TL;DR

This systematic review and meta-analysis of 34 quasi-experimental studies found that childhood maltreatment remained associated with mental health problems after stringent adjustment for confounding, consistent with a small causal contribution. The adjusted association was substantially smaller than the unadjusted association, suggesting that part of the overall risk reflects wider genetic and environmental risk factors.

Key findings

  • Across 34 quasi-experimental studies including 54,646 independent participants, childhood maltreatment was associated with mental health problems after quasi-experimental adjustment (Cohen’s d=0.31, 95% CI=0.24–0.37, I²=76.27).
  • Before quasi-experimental adjustment, the association was moderate (Cohen’s d=0.56, 95% CI=0.41–0.71, I²=97.29); the adjusted estimate was 45% smaller, and the difference was significant (p=0.004).
  • The adjusted association was consistent across quasi-experimental methods (Q_moderation=3.43, p=0.75) and generalized across psychiatric disorders, but was moderated by maltreatment type (Q_moderation=22.43, p=0.0042), with emotional abuse and institutional neglect more strongly associated than some other subtypes/composite measures; these subtype findings were based on limited data (emotional abuse: 3 studies, 7 effect sizes; institutional neglect: ERA Study only).
  • Sensitivity analyses suggested small-study bias (Egger’s test p=0.004) attributable to the ERA cohort; after excluding ERA, p=0.09. Leave-one-out estimates ranged from Cohen’s d=0.29 to 0.32, and excluding broader victimization/ACEs or institutional neglect studies yielded Cohen’s d=0.34, 95% CI=0.26–0.42.

Methods

  • Pre-registered systematic review and meta-analysis (PROSPERO CRD42020187520), following PRISMA and MOOSE; searched Embase, PsycINFO, and MEDLINE from inception until January 1, 2022 for peer-reviewed English-language articles.
  • Included 34 quasi-experimental studies based on 29 distinct cohorts, comprising 54,646 participants in adjusted analyses (56.72% female; mean age 28.2 years at mental health assessment), yielding 156 adjusted and 103 unadjusted effect sizes.
  • Used multilevel random-effects meta-analysis in R with metafor and dmetar, converting effect sizes to Cohen’s d, applying robust standard errors, estimating heterogeneity with I², and conducting sensitivity analyses (Egger’s test, p-curve, leave-one-out, definition robustness) plus meta-regression for moderators.

Limitations

  • The paper states that each quasi-experimental approach has potential biases (e.g., family-based designs do not account for non-familial confounding; panel designs can be affected by time-varying confounding; propensity score approaches are liable to unmeasured confounding; natural experiments may not examine generalizable exposures), so unmeasured confounding cannot be entirely ruled out despite triangulation.
  • The paper states that method type was related to maltreatment measurement, that it could not disentangle these in multivariate moderation due limited data, that specific maltreatment-type effects cannot be firmly concluded because poly-victimization is common and rarely controlled, and that data were lacking on timing of maltreatment, time interval to psychopathology, and race or ethnicity.
  • Evident from the text: adjusted analyses showed high heterogeneity (I²=76.27), Egger’s test indicated small-study bias driven by the ERA cohort, and 25 of 34 studies did not report race or ethnicity, limiting moderation by race/ethnicity.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1176/appi.ajp.20220174

Cites

84 of 84 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 Forest plot depicting the study-average effects of child maltreatment on mental health from quasi-experimental studies. Note. MREM = multi-level random-effects meta-analysis model. For clarity of presentation, the forest plot shows 1 effect size per study (reflecting the average of all individual effect sizes obtained from each study), rather than all 150 effect sizes used to derive the pooled MREM estimate. The average effect size per study and its variance was calculated using the MAd package,(87) assuming a correlation of 0.6 between multiple within-study effect sizes. The 156 effect sizes are presented in Table S9 of the Supplement. I2 for the MREM was 76.27, indicating that 76% of variation between effect sizes would remain if sampling error was eliminated.
Figure 1 Forest plot depicting the study-average effects of child maltreatment on mental health from quasi-experimental studies. Note. MREM = multi-level random-effects meta-analysis model. For clarity of presentation, the forest plot shows 1 effect size per study (reflecting the average of all individual effect sizes obtained from each study), rather than all 150 effect sizes used to derive the pooled MREM estimate. The average effect size per study and its variance was calculated using the MAd package,(87) assuming a correlation of 0.6 between multiple within-study effect sizes. The 156 effect sizes are presented in Table S9 of the Supplement. I2 for the MREM was 76.27, indicating that 76% of variation between effect sizes would remain if sampling error was eliminated.CC BY · PubMed Central
Figure 2 Meta-analytic associations between child maltreatment and mental health problems across different quasi-experimental methods. Note. k=number of studies (i.e., papers); n=number of participants across studies; ES=number of effect sizes. “MZ twin difference” refers to twin difference designs including only monozygotic twins, while “twin difference” refers to twin difference designs that include both monozygotic and dizygotic twins. Quasi-experimental methods used by only a single study in the meta-analysis (namely the adoption design (43), Children of Twins design (42), and random-intercept cross-lagged panel model (55)) were not included in the moderator analysis.
Figure 2 Meta-analytic associations between child maltreatment and mental health problems across different quasi-experimental methods. Note. k=number of studies (i.e., papers); n=number of participants across studies; ES=number of effect sizes. “MZ twin difference” refers to twin difference designs including only monozygotic twins, while “twin difference” refers to twin difference designs that include both monozygotic and dizygotic twins. Quasi-experimental methods used by only a single study in the meta-analysis (namely the adoption design (43), Children of Twins design (42), and random-intercept cross-lagged panel model (55)) were not included in the moderator analysis.CC BY · PubMed Central
Figure 3 Meta-analytic associations between child maltreatment and different mental health problems Note. k=number of studies (i.e., papers); n=number of participants across studies; ES=number of effect sizes; ADHD=attention-deficit/hyperactivity disorder. Specific mental health outcomes were grouped into categories shown in the figure. Internalising problems includes internalising symptoms, internalising behavior, emotional symptoms, and trauma symptoms; conduct problems includes conduct disorder, conduct problems, antisocial, oppositional, or aggressive behavior, arrest or incarceration; externalising problems includes externalising symptoms or behavior; personality disorder includes borderline, paranoid, schizoid, schizotypal, histrionic, narcissistic, antisocial, avoidant, obsessive-compulsive and dependent personality disorders; psychopathology broad includes total psychopathology symptoms, the p-factor, any psychopathology disorder, and total behaviour problems (on the Strengths and Difficulties Questionnaire). Outcomes assessed in only one study (bulimia (37) and substance use disorder (54)) were not included in the moderator analysis.
Figure 3 Meta-analytic associations between child maltreatment and different mental health problems Note. k=number of studies (i.e., papers); n=number of participants across studies; ES=number of effect sizes; ADHD=attention-deficit/hyperactivity disorder. Specific mental health outcomes were grouped into categories shown in the figure. Internalising problems includes internalising symptoms, internalising behavior, emotional symptoms, and trauma symptoms; conduct problems includes conduct disorder, conduct problems, antisocial, oppositional, or aggressive behavior, arrest or incarceration; externalising problems includes externalising symptoms or behavior; personality disorder includes borderline, paranoid, schizoid, schizotypal, histrionic, narcissistic, antisocial, avoidant, obsessive-compulsive and dependent personality disorders; psychopathology broad includes total psychopathology symptoms, the p-factor, any psychopathology disorder, and total behaviour problems (on the Strengths and Difficulties Questionnaire). Outcomes assessed in only one study (bulimia (37) and substance use disorder (54)) were not included in the moderator analysis.CC BY · PubMed Central
Figure 4 Meta-analytic associations between different types of child maltreatment and mental health problems Note. k=number of studies (i.e., papers); n=number of participants across studies; ES=number of effect sizes. Maltreatment includes assessment of multiple subtypes of abuse and/or neglect; victimisation includes assessment of maltreatment alongside other forms of victimisation (e.g., bullying); ACEs includes assessment of maltreatment alongside other forms of adverse childhood experiences.
Figure 4 Meta-analytic associations between different types of child maltreatment and mental health problems Note. k=number of studies (i.e., papers); n=number of participants across studies; ES=number of effect sizes. Maltreatment includes assessment of multiple subtypes of abuse and/or neglect; victimisation includes assessment of maltreatment alongside other forms of victimisation (e.g., bullying); ACEs includes assessment of maltreatment alongside other forms of adverse childhood experiences.CC BY · PubMed Central

Verification

Claims the paper makes

13 Valid · 5 Not valid · 5 Not checked · 65 Own results and interpretations

57 more claims · 6 with their cited sources checked

Create a free account to see all 88 claims and to verify the 6 references still open.

See all claims — free account
  • It may state two separate claims as one. 17
  • Probably not a substantial claim. 14

Create a free account to see these claims.

Citations

Cites

Show all 84 references

Research works printed but not linked: 3

  1. Langevin R, Trauma Violence Abuse
  2. Baldwin JR, Nat Hum Behav
  3. Kullberg M-L, 2020, Psychol Med, 51, 1

Cited by

Items 101–200 of 306

Metadata

Authors
Baldwin, J. R., Wang, B., Karwatowska, L., Schoeler, T., Tsaligopoulou, A., Munafò, M. R., & Pingault, J.-B.
Affiliations
  • King's College London (GB)
  • Department of Clinical, Educational, and Health Psychology, Division of Psychology and Language Sciences, University College London (Baldwin, Wang, Schoeler, Tsaligopoulou, Pingault); Social, Genetic, and Developmental Psychiatry Centre, Institute of Psychiatry, Psychology, and Neuroscience, King’s College London (Baldwin, Pingault); Great Ormond Street Institute of Child Health, University College London (Karwatowska); Department of Computational Biology, University of Lausanne, Lausanne, Switzerland ...
Corresponding author
Baldwin, Jessie R.
Venue
American Journal of Psychiatry
Published
2023-02-01
Volume
180
Issue
2
Pages
117-126
ISSN
  • 0002-953X (ISSN-L)
  • 1535-7228
OpenAlex
W4315619289
Cited by
313
References
84
Citations by year
2026: 125 · 2025: 104 · 2024: 56 · 2023: 26
FWCI
15.77
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • American Journal of Psychiatry (publishedVersion)
  • PubMed (publishedVersion)
  • PubMed Central (acceptedVersion, cc-by)
  • SERVAL (Université de Lausanne) (acceptedVersion, cc-by)
  • Bristol Research (University of Bristol) (acceptedVersion, cc-by)
  • Explore Bristol Research (acceptedVersion, cc-by)
  • UCL Discovery (University College London) (submittedVersion)
  • IRIS (submittedVersion, cc-by)
Funders
  • Wellcome Trust (215917/Z/19/Z, 215917, 218641, 218641/Z/19/Z)
  • European Commission (863981, HORIZON2020)
  • Medical Research Council (MC_UU_00011/7, MC_UU_00011)
Topics
  • Social Sciences › Psychology › Clinical Psychology › Child Abuse and Trauma
  • Social Sciences › Sociology and Political Science › Child Welfare and Adoption
  • Social Sciences › Psychology › Clinical Psychology › Child and Adolescent Psychosocial and Emotional Development
Keywords
child abuse, mental illness, causal inference, quasi-experimental design, genetic and environmental risk factors, psychopathology prevention
MeSH terms
Mental Disorders, Child, Child Abuse, Humans, Mental Health, Psychopathology, Twins

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.