Potentially modifiable mediators of the association between child abuse and dementia

Technical Summary: Potentially Modifiable Mediators of the Association Between Child Abuse and Dementia

Problem Statement
Child abuse is a known risk factor for dementia, associated with a significantly increased hazard ratio. However, existing literature investigating the causal pathways between child abuse and dementia faces several limitations. First, many studies utilize composite Adverse Childhood Experience (ACE) scores, which assume uniform impacts across different types of adversity, potentially diluting the specific association between child abuse and dementia. Second, prior research often examines mediators (e.g., socioeconomic status, health behaviors, mental health) in isolation, failing to account for the interdependencies and temporal sequences among multiple mediators. Third, much of the existing evidence relies on the UK Biobank (UKB), a cohort that may lack generalizability to more vulnerable, lower socioeconomic populations. Finally, methodological approaches in previous studies often rely on counterfactual frameworks that require strong assumptions regarding unmeasured confounding and intermediate confounders, which can introduce collider bias.

Methodology
This study utilized data from the English Longitudinal Study of Ageing (ELSA), a nationally representative cohort of adults aged 50 and older, to investigate mediation pathways. The analytical sample consisted of 5,448 individuals who completed a retrospective life history interview regarding child abuse.

• Exposure: Child abuse was defined retrospectively as experiencing sexual assault, serious physical attack, or physically abusive parents before age 16.
• Outcome: All-cause dementia was defined algorithmically based on DSM-IV criteria, incorporating self-reported diagnoses, informant questionnaires (IQCODE), and cognitive decline metrics.
• Mediators: Four categories of potentially modifiable risk factors were selected based on the 2024 Lancet Commission and evidence linking them to child abuse:
  1. Education: Years of education.
  2. Health Behaviors: Smoking, alcohol consumption, and physical inactivity.
  3. Mental Health: Depression and anxiety.
  4. Cardiovascular Health: Hypertension, obesity, diabetes, and cholesterol.
• Statistical Approach: The study employed causal mediation analysis using the g-formula to estimate interventional effects. This approach allowed for the joint modeling of multiple correlated mediators without requiring the strict temporal ordering assumptions often violated in single-mediator analyses. The analysis estimated the Randomised Total Effect (RTE), Randomised Natural Direct Effect (RNDE), and Randomised Natural Indirect Effect (RNIE).
• Adjustments: Models were adjusted for sex, age, age squared, and childhood socioeconomic status (SES). Multiple imputation was used to handle missing data, and sensitivity analyses were conducted using complete case data and alternative model specifications.

Key Results

• Association: Individuals who experienced child abuse had an 80% higher hazard of developing dementia compared to those who did not (RTE HR: 1.80, 95% CI: 1.21–2.39).
• Mediation Analysis: The four mediator categories (education, health behaviors, mental health, and cardiovascular health) jointly mediated approximately 18% of the association between child abuse and dementia.
• Direct vs. Indirect Effects: The Randomised Natural Direct Effect (RNDE) was 1.66 (95% CI: 1.12–2.20), indicating that the majority of the association remains unexplained by these modifiable factors.
• Specific Mediator Contributions:
  • Mental Health: This was the primary driver of mediation. Child abuse was strongly associated with depression (OR: 2.24) and anxiety (OR: 2.90), which in turn were associated with higher dementia hazards. Sensitivity analyses suggested that mental health accounted for nearly all of the observed 18% mediation (approximately 15–17%).
  • Other Factors: Evidence for mediation through education, health behaviors, and cardiovascular health was weak. Notably, no significant association was found between child abuse and years of education in this cohort, nor between education and dementia incidence. Associations between child abuse and cardiovascular risk factors were imprecise and generally non-significant.
• Interventional Estimate: A joint intervention on all four mediator categories among those exposed to child abuse was estimated to reduce their dementia hazard by approximately 8% (RNIE: 1.08).

Significance and Claims
The authors claim that this study provides a more nuanced understanding of the pathways from child abuse to dementia by moving beyond composite ACE scores and single-mediator analyses. By using the g-formula on a representative UK cohort, the study demonstrates that while child abuse is a robust predictor of dementia, currently known modifiable risk factors (education, lifestyle, cardiovascular health, and mental health) explain only a modest proportion of this risk.

The paper emphasizes that mental health is the dominant mediator among the factors studied. Consequently, the authors argue that current dementia prevention strategies, which focus heavily on cardiovascular risk and lifestyle modification in later life, are unlikely to substantially reduce the excess dementia risk specifically attributable to child abuse. The study concludes that future research must prioritize identifying other pathways, including biological mechanisms (e.g., inflammation, neuroendocrine dysregulation) and psychosocial factors not captured in the dataset (e.g., PTSD symptoms, adult adversity), to develop effective secondary prevention interventions for this vulnerable population.