Dube · JAMA 2001 · retrospective cohort study · n=17,337

Childhood abuse, household dysfunction, and the risk of attempted suicide throughout the life span: findings from the Adverse Childhood Experiences Study.

Cited 2312 times in the scientific literature.

Level 3 - non-randomized controlled study

Retrospective cohort study using survey data from an adult HMO population

PubMed 11754674 · doi:10.1001/jama.286.24.3089 · record verified 2026-08-31

What was done

A retrospective cohort study was conducted among 17,337 adult HMO members (54% female; mean age 57 years) visiting a primary care clinic in San Diego between 1995 and 1997. Participants completed a survey assessing adverse childhood experiences (ACEs)—including abuse (emotional, physical, sexual) and household dysfunction (substance abuse, mental illness, incarceration, domestic violence, parental separation/divorce)—as well as self-reported history of suicide attempts and age at first attempt.

What was found

The lifetime prevalence of suicide attempts was 3.8% overall, ranging from 1.1% among individuals with 0 ACEs to 35.2% among those with 7 or more ACEs. Each ACE category increased suicide attempt risk 2- to 5-fold. The ACE score demonstrated a graded dose-response relationship with suicide attempts (P < .001); individuals with ≥7 ACEs had an adjusted odds ratio of 31.1 (95% CI, 20.6–47.1) for ever attempting suicide compared to those with 0 ACEs. Adjustment for illicit drug use, depressed affect, and alcoholism attenuated the association, indicating partial mediation. Population-attributable risk fractions for ≥1 ACE were 67% for lifetime, 64% for adult, and 80% for childhood/adolescent suicide attempts.

Why it matters

This study demonstrates a massive, graded relationship between cumulative childhood trauma and lifelong suicide attempts, highlighting that early-life adversity is a major contributor to suicidal behavior.

Limits

The study relied on retrospective self-report for both childhood exposures and suicide attempts, introducing potential recall and reporting bias. The sample was restricted to insured HMO members attending primary care, which may limit generalizability to uninsured or institutionalized populations. Non-fatal suicide attempts were assessed, so completed suicides were not captured.

Cited by