Research · Threat physiology · October 7, 2026

ACE scores predict populations, not people

●●●○Evidence: Supported

Childhood adversity raises risk across a population. As a forecast for any one person, an ACE score performs only modestly better than chance.

What the research found

The original ACE Study found a graded relationship: the more kinds of childhood adversity someone reported, the higher their adult risk for a long list of health problems. A meta-analysis of 37 studies and more than 250,000 people put numbers on it. Compared with people reporting no ACEs, people with four or more had about 4.4 times the odds of depression and about 10 times the odds of problematic drug use.

A 2021 study asked a different question: can an ACE score predict which individuals will develop problems? Following two birth cohorts, one in the UK and one in New Zealand, the answer was “barely.” The scores separated people who later had health problems from those who didn’t only modestly better than chance (AUCs around 0.57 to 0.63). In one cohort, 56% of children with four or more ACEs had a mental health problem by 18, against 33% of children with none.

How strong is this?

Supported. The population-level link is one of the most replicated findings in public health, though it rests on observational data and adult recall. The individual-prediction finding comes from two well-run prospective cohorts.

What it does and doesn’t mean

Both things are true at once. ACEs change the odds across a population, and an ACE score is a weak crystal ball for a single life. That matters for how we screen and how we talk to people about their histories. A high score tells a clinician where to look. It doesn’t tell anyone what they’ll find, or set a person’s ceiling.

Sources

1. Felitti et al. 1998, American Journal of Preventive Medicine 14(4):245–258
2. Hughes et al. 2017, Lancet Public Health 2(8):e356–e366
3. Baldwin et al. 2021, JAMA Pediatrics 175(4):385–393