Annual Research Review: Umbrella synthesis of meta-analyses on child maltreatment antecedents and interventions: differential susceptibility perspective on risk and resilience

Abstract

Child maltreatment in the family context is a prevalent and pervasive phenomenon in many modern societies. The global perpetration of child abuse and neglect stands in stark contrast to its almost universal condemnation as exemplified in the United Nation's Convention on the Rights of the Child. Much work has been devoted to the task of prevention, yet a grand synthesis of the literature is missing. Focusing on two core elements of prevention, that is, antecedents for maltreatment and the effectiveness of (preventative) interventions, we performed an umbrella review of meta-analyses published between January 1, 2014, and December 17, 2018. Meta-analyses were systematically collected, assessed, and integrated following a uniform approach to allow their comparison across domains. From this analysis of thousands of studies including almost 1.5 million participants, the following risk factors were derived: parental experience of maltreatment in his or her own childhood (d = .47), low socioeconomic status of the family (d = .34), dependent and aggressive parental personality (d = .45), intimate partner violence (d = .41), and higher baseline autonomic nervous system activity (d = .24). The effect size for autonomic stress reactivity was not significant (d = -.10). The umbrella review of interventions to prevent or reduce child maltreatment showed modest intervention effectiveness (d = .23 for interventions targeting child abuse potential or families with self-reported maltreatment and d = .27 for officially reported child maltreatment cases). Despite numerous studies on child maltreatment, some large gaps in our knowledge of antecedents exist. Neurobiological antecedents should receive more research investment. Differential susceptibility theory may shed more light on questions aimed at breaking the intergenerational transmission of maltreatment and on the modest (preventive) intervention effects. In combination with family-based interaction-focused interventions, large-scale socioeconomic experiments such as cash transfer trials and experiments with vouchers to move to a lower-poverty area might be tested to prevent or reduce child maltreatment. Prevalence, antecedents, and preventive interventions of prenatal maltreatment deserve continuing scientific, clinical, and policy attention.

Keywords: Child maltreatment; interventions; meta-analysis; umbrella synthesis.

Figure 1

Risk and protective factors for child maltreatment. Note: Heuristic model of risk and protective factors for child maltreatment, with socioeconomic adversities and intergenerational experiences as more distal factors that are suggested to influence type and chronicity of maltreatment, mediated by more proximal factors on the psychological and neurobiological level. Distal and proximal influences are moderated by parent or child protective factors (constitutional characteristics, e.g., easy temperament) and protective factors in the social context (e.g., parent coaching intervention) as predicted by differential susceptibility theory. [Colour figure can be viewed at http://www.wileyonlinelibrary.com]

Figure 2

PRISMA flowchart of included and excluded studies. *Corrections to included papers (k = 1); TI/AB = Title/Abstract screening. [Colour figure can be viewed at http://www.wileyonlinelibrary.com]

Figure 3

Umbrella effect sizes (d) for antecedents of child maltreatment and effectiveness of interventions to reduce or prevent maltreatment. Note: In this Figure, the effect sizes (Cohen’s d) combined across several meta‐analyses are presented, starting with parents having been victim of maltreatment in their own childhood, parental personality influences, having been victim of interpersonal or intimate partner violence (IPV), low socioeconomic status (SES), and baseline regulation of the autonomic nervous system (ANS). The final two antecedents are the results of interventions focusing on child maltreatment itself, and interventions targeting individuals with elevated risk of becoming a perpetrator (‘child abuse potential’ or CAP). All combined effect sizes are significant and in the expected direction. CM = Child maltreatment; IPV = Intimate partner violence; SES = Socioeconomic status; ANS = Autonomic nervous system; CAP = Child abuse potential. The effect size for ANS reactivity was d = −.10 and nonsignificant. [Colour figure can be viewed at http://www.wileyonlinelibrary.com]

Figure 4

A differential susceptibility perspective on risk and resilience. Note: Resilient individuals show lower susceptibility to good as well as bad environments. More resilient, less susceptible parents might persist in ‘good‐enough’ parenting even in adverse circumstances, and more resilient, less susceptible children may survive maltreatment with fewer scars than their more susceptible but less resilient peers. The less susceptible individuals can rely on their constitutional or high internal resilience rooted in a less reactive temperament and neurobiological makeup, whereas the more susceptible individuals must rely more on a supportive environment as a buffer against adversities – which is high external resilience [Colour figure can be viewed at http://www.wileyonlinelibrary.com]