The association between childhood maltreatment and conduct disorder: a systematic review and three level random effects meta-analysis

Background

Conduct disorder (CD) is a severe and heterogeneous psychiatric condition of childhood and adolescence, characterised by a persistent pattern of aggression, rule breaking, deceitfulness, and violations of the rights of others [1]. It affects approximately 3% of young people worldwide and represents a major contributor to the global burden of mental disorders [2]. CD is associated with substantial short- and long-term adverse outcomes, including interpersonal difficulties, involvement with the youth criminal justice system, poor educational attainment, substance misuse, and elevated rates of psychiatric comorbidity. These outcomes contribute to significant individual, familial, and societal costs across the life course.

Presentations of CD differ markedly between individuals, reflecting complex interactions between biological vulnerabilities and environmental exposures. Childhood maltreatment (CM), including experiences of abuse and neglect, is one of the most consistently identified environmental risk factors for CD [3]. However, the existing literature is fragmented, with considerable variation in study design, sample characteristics, assessment methods, and definitions of maltreatment. As a result, the strength and specificity of associations between CM and CD remain unclear and it is not well understood how methodological and demographic factors influence reported associations.

Aims

This systematic review and meta-analysis aimed to: (a) quantify the overall association between childhood maltreatment and conduct disorder in youth, (b) examine whether the magnitude of this association differs between maltreatment subtypes, including physical abuse, emotional abuse, sexual abuse, physical neglect, emotional neglect, and witnessing domestic violence and (c) explore whether study level, sample level, and measurement level moderators influence effect size variability.

Methods

Eligible studies assessed both CM and CD occurring before the age of 18 years. A systematic search of six electronic databases identified 33 eligible studies, comprising 58% cross sectional and 42% longitudinal designs. These studies contributed 74 independent effect sizes, representing a total of 143,422 participants. A multilevel random effects meta analysis was conducted using restricted maximum likelihood estimation to appropriately account for statistical dependencies between effect sizes. All analyses were performed in R [4] using the metafor package [5].

Results

A large and statistically significant positive association was observed between CM and CD (r = 0.31, 95% CI [0.24, 0.38], p < .001). Significant between study heterogeneity was present (Q = 475.49, p < .001), with a moderate proportion of variance attributable to true differences between studies (I² = 48%). Moderator analyses indicated that CM subtype did not significantly account for this heterogeneity and pairwise comparisons confirmed that the magnitude of association between CM and CD did not differ significantly across maltreatment subtypes.

Conclusions

CM is robustly associated with CD in youth, regardless of maltreatment subtype. Formal assessments of study quality, publication bias, and sensitivity are ongoing and will be reported in subsequent analyses. While moderate heterogeneity was observed between studies, ongoing multiple‑moderator analyses aim to further clarify the contribution of study design, sample characteristics, and measurement factors to this variability. Improved understanding of these sources of heterogeneity may help inform the development of targeted prevention, early identification and better interventions for children and adolescents.

References

1. American Psychiatric Association, 2013. Diagnostic and statistical manual of mental disorders, fifth ed. American Psychiatric Association Publishing, Washington.
2. Tian, J., Yan, N., Hu, X., et al., 2025. Global burden of mental disorders among adolescents and young adults, 1990–2021: a systematic analysis of the Global Burden of Diseases Study 2021. General Psychiatry 38, e102278. http://dx.doi.org/10.1136/gpsych-2025-102278.
3. Fairchild, G., Hawes, D.J., Frick, P.J., Copeland, W.E., Odgers, C.L., Franke, B., Freitag, C.M., De Brito, S.A., 2019. Conduct disorder. Nat. Rev. Dis. Primers 5 (1), 43. http://dx.doi.org/10.1038/s41572-019-0095-y.
4. R Core Team, 2024. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Available from: http://www.r-project.org/.
5. Viechtbauer, W., 2010. Conducting meta-analyses in R with the metafor package. J. Stat. Softw. 36 (3), 1–48. http://dx.doi.org/10.18637/jss.v036.i03.