Why Prevention Targets the Young

February 11, 2024

Most money spent on mental health goes toward treating adults who are already unwell. Yet a growing share of prevention funding, screening effort, and research attention is being redirected toward children and adolescents, years before most of them show any sign of a problem. This is not a fad or a marketing angle for a new generation of apps. It actually makes sense. It is a response to three separate lines of evidence that converge on the same conclusion: childhood and adolescence are the period in which most of the lifetime burden of mental illness is either created or headed off, and the expected return on intervening early is structurally different from the return on intervening later.

The Adverse Childhood Experiences Evidence

The single most influential data point behind the youth focus is not really about mental health at all in its original framing. In 1998, Vincent Felitti and Robert Anda published results from more than 17,000 adult members of a California health plan who had answered a questionnaire about seven categories of adversity experienced before age 18: physical, sexual, and emotional abuse; physical and emotional neglect; and growing up in a household with substance abuse, mental illness, incarceration, or domestic violence. Each category present in a person’s childhood added one point to an “ACE score,” and the study found a strikingly clean dose-response relationship between that score and dozens of adult health outcomes, not just psychiatric ones but heart disease, cancer, and liver disease as well. One caveat here is that ACE tends to be correlated to many other risk factors such as socio-economic status (SES).

A follow-up analysis of the same cohort found that, compared with someone with an ACE score of zero, someone with a score of four or more had roughly twelve times the odds of having ever attempted suicide. The original 1998 paper reported four- to twelve-fold increases across depression, alcoholism, drug use, and suicide attempts for the same comparison. These are not subtle effects sitting at the edge of statistical noise. They are among the largest, most reproducible dose-response relationships in all of psychiatric epidemiology, and they have since been replicated in dozens of countries and cohorts.

The mechanism is not mysterious. Chronic, unbuffered stress in childhood, what developmental scientists call toxic stress, keeps the hypothalamic-pituitary-adrenal axis activated for extended periods during a stage when the brain and endocrine system are still being built rather than merely used. Cortisol exposure during this period shapes the developing amygdala, hippocampus, and prefrontal cortex in ways that persist into adulthood as altered stress reactivity, impaired emotion regulation, and, in a meaningful fraction of cases, a diagnosable disorder. Childhood experiences have outsized impacts on developing brains. For example, if people that were supposed to protect you instead betray you, that updates a child’s belief systems significantly, perhaps internalizing that no one can be trusted. Preventing an adverse childhood experience is therefore not just a moral good in its own right. It removes one of the strongest known upstream causes of adult mental illness, which is why so much of prevention science has quietly become childhood policy: home visiting programs, parenting support, poverty reduction, and school-based safety, more than anything delivered inside a therapy room.

Adolescence as a Biological Window

Childhood adversity is only half of the argument. The other half is that adolescence itself is a distinct neurobiological window, not just a convenient administrative category. The prefrontal cortex, the region most responsible for emotion regulation, planning, and impulse control, does not finish maturing until the mid-twenties. Synaptic density in this region peaks in late childhood and is then pruned across the second and third decades of life, a period during which the circuit is unusually plastic and unusually exposed. Researchers who study this window, Sarah-Jayne Blakemore prominent among them, describe adolescence as a sensitive period in the same technical sense used for early language acquisition: a stretch of development in which experience has an outsized and comparatively permanent effect on the resulting architecture.

This plasticity cuts both ways. It is what makes early intervention unusually effective, because a nervous system still under construction responds to treatment, skill-building, and environmental change in ways a fully matured one does not. It is also what makes this the period in which untreated stress and illness leave the deepest mark, because the same openness that allows benefit also allows harm to become structural rather than transient.

There is a second, more specific mechanism worth naming: kindling. In 1992, Robert Post proposed that recurrent mood episodes are not independent events but a self-reinforcing process. A first depressive or manic episode is typically triggered by a significant external stressor. But the episode itself appears to leave a biological trace, changes in gene expression and neural circuitry, that lowers the threshold required to trigger the next one. Later episodes increasingly arrive with less external provocation, sometimes with none at all. Subsequent work by Lars Kessing and colleagues, following patients across their full clinical course, confirmed that the risk of recurrence rises with each additional episode a person has already had. The practical implication is blunt: the first episode is not just the first data point in a series, it is the event that makes the rest of the series more likely.

This is the part of the argument that gets lost when the age focus is described only as “catching problems early.” It is not just earlier detection of a disease that would have arrived anyway. For a meaningful fraction of people, whether the first episode happens at all is the pivot point, and once it happens the odds shift, sometimes permanently, toward a recurrent course.

When Onset Happens, and What That Predicts

If adolescence is a genuine biological window, the epidemiology should show it, and it does. A 2022 meta-analysis by Marco Solmi and colleagues, pooling 192 studies and more than 700,000 participants worldwide, found that even the latest-onset major category of disorder, mood disorders, has a median age of onset in the early thirties, while neurodevelopmental disorders, anxiety, and schizophrenia-spectrum disorders all cluster earlier still. Ronald Kessler’s earlier and now-canonical estimate, drawn from a nationally representative US sample, put it in a single memorable number: half of all mental disorders that a person will ever experience have already begun by age 14, and three-quarters by age 24. A 2023 cross-national study spanning 29 countries and more than 150,000 respondents, led by John McGrath, found the same pattern outside the US: the single most common age for a first episode of any disorder was 15.

The other half of this picture, the part that gets less attention, is what these curves imply about people who pass through adolescence and their twenties without a first episode. Incidence is not flat across the lifespan. It is heavily front-loaded, which means the annual probability of a first onset keeps falling the further someone gets past the window without one. Depression in particular follows a roughly bimodal pattern: a large early-adulthood peak and a much smaller peak late in life. The late peak, moreover, is not simply the same disease arriving on a delay. Late-onset depression is disproportionately associated with cerebrovascular disease, has a weaker family-history signal, and responds differently to treatment, findings that underpin what geriatric psychiatrists call the vascular depression hypothesis. In practice, this means a person who reaches midlife without ever having had a mood episode is not a ticking clock waiting to go off. Their risk has genuinely fallen, and if a first episode does eventually arrive, it is disproportionately likely to be a different kind of illness with a different cause than the one prevention programs in adolescence are built to interrupt.

The Utilitarian Case

Even without any of the mechanistic argument above, there is a separate and simpler case for the youth focus, one that has nothing to do with prevention being more effective in adolescence and everything to do with what is being purchased with a fixed unit of resource. Health economists measure this with quality-adjusted life years: a treatment that prevents a chronic, recurrent illness in a sixteen-year-old is buying up to seven decades of averted disability, while the same intervention delivered to someone who develops a first episode at seventy is buying, at most, a handful of years. Alan Williams formalized this reasoning as the fair innings argument: fairness, in his framing, requires that people get a reasonable share of healthy life before resources are directed elsewhere, so an early illness that threatens to shorten someone’s innings carries a stronger moral claim than an equivalent illness arriving after a full life has already been lived.

The kindling evidence above sharpens this further. A prevented first episode in an adolescent is not one averted illness, it is an averted chain, because each episode a person does not have is also an episode that cannot lower the threshold for the next one. The expected value of intervening before the first episode compounds in a way that intervening after the fifth episode does not.

What the Focus on Youth Gets Wrong

None of this is a case for treating the age target as unconditionally correct. It has real costs, and they deserve as much weight as the evidence above.

The clearest is overdiagnosis. Lucy Foulkes and Jack Andrews have documented what they call the prevalence inflation hypothesis: mental health awareness campaigns aimed at young people appear to be increasing reported rates of distress partly through genuine improved recognition, but partly through overinterpretation, in which normal adolescent moodiness, a developmentally appropriate and largely self-limiting state, gets relabeled as a disorder. Screening a population at the exact age when identity, mood, and behavior are naturally in flux runs a structural risk of false positives that screening a stable adult population does not. Emerging evidence shows that mindfulness courses in schools can negatively affect those who are otherwise well.

Labeling has its own cost independent of accuracy. A psychiatric diagnosis attached during adolescence, a period when identity is still being formed, can become part of how a young person understands themselves in a way that a diagnosis received at forty rarely does. Some of this labeling effect helps, by giving people language and access to care. Some of it plausibly harms, by narrowing a still-forming sense of self around a clinical category before that category has proven itself accurate.

There is also a resource allocation cost that the fair innings argument tends to understate. Directing prevention funding, screening infrastructure, and specialist attention toward the young pulls the same resources away from adults with established, severe, and often more treatable-in-the-moment illness. The rule of rescue, the strong intuition that a person suffering acutely in front of you has a claim on help that a statistical future case does not, cuts directly against the actuarial logic of the youth focus. A society optimizing purely for life-years averted risks deprioritizing exactly the people whose suffering is most visible and immediate.

Children do not exist in a vacuum, and struggling children often have struggling parents. Any approach that only targets children may be ineffective. A more holistic approach would target the entire world that the child lives in.

Finally, much of the evidence for the mechanisms above is stronger than the evidence for the universal prevention programs built on top of them. ACE scores predict outcomes robustly; that does not mean any given school-based prevention curriculum reduces those outcomes at scale, and several large trials of universal youth mental health programs have produced disappointing or null results. The age-of-onset data is a strong argument for where to look. It is a much weaker argument for any specific thing to do once you are looking there.

Conclusion

The case for targeting youth in mental health is unusually well evidenced by the standards of the field. Childhood adversity is one of the most reproducible risk factors in psychiatric epidemiology. Adolescence is a genuine, time-limited window of neurobiological plasticity. Most disorders begin before most people finish their education, and the ones that begin later are often different in kind, not just in timing. Layered on top of the biology, the utilitarian arithmetic of life-years is not close. None of this means the focus is free of cost. It carries real risks of overdiagnosis, premature labeling, and resource diversion away from people who are suffering acutely right now. Taking the youth argument seriously means taking those costs seriously too, not treating an evidence base for where the leverage is as a license for how heavily to intervene.