Does early-age alcohol initiation raise mental disorder risk in children and adolescents?

A two-year test of timing, not just amount

Most work linking alcohol to poor psychiatric outcomes has looked at adolescents or adults, or has not established which came first, drinking or symptoms. This study asked a narrower question: in children who were free of mental disorders at baseline, does early alcohol experimentation predict new clinical diagnoses within a short window?

The authors used the Adolescent Brain Cognitive Development (ABCD) study, a U.S. cohort of 11,878 children recruited at ages 9-10 across 21 institutions. Substance use was collected by semi-structured interview at baseline and at 6, 12, 18 and 24 months. A child counted as an alcohol experimenter if they reported at least one sip of alcohol, whether before baseline or during follow-up.

After excluding children with other substance use, prior mental disorders, traumatic brain injury, severe physical illness, missing drinking status, or incomplete two-year follow-up, the authors matched each experimenter to a child with no alcohol use on age and sex. The analyzed sample was 1,659 experimenters and 1,659 controls (n=3,318). Mean baseline age was 9.55 years (SD 0.50), and 944 experimenters (56.9%) were male.

Diagnoses came from clinical interviews, not questionnaires

The outcome was not a symptom score. Mental disorders were diagnosed by professional psychiatrists using the Kiddie Schedule for Affective Disorders and Schizophrenia, Present and Lifetime Version (K-SADS-PL), aligned with DSM-5, assessed at baseline, 12 months and 24 months. Restricting the analysis to children without a baseline diagnosis is what gives the design its temporal ordering: the diagnosis had to appear after the reported sipping.

Exposure was measured in two ways. The first was a simple yes/no: had the child ever taken at least one sip. The second used a latent class trajectory model (LCTM), a mixed-modeling approach for repeated measurements, to group experimenters by how their sipping status changed across the four follow-up points. That second step matters because a single yes/no flag cannot distinguish a child who sipped once from one whose drinking was escalating.

Odds ratios were adjusted for participant demographics (age, sex at birth, race and ethnicity, education level), caregiver characteristics (age, education, marital status, depression and anxiety symptoms) and family history of mental disorders.

A modest overall increase in new diagnoses

Across the full matched sample, early alcohol initiation was associated with higher odds of developing a mental disorder: OR 1.25 (95% CI 1.03-1.52). The confidence interval sits close to 1, so the overall estimate is compatible with a small effect.

The paper's discussion describes this as a 25% increased risk. That phrasing refers to odds in a logistic model, not to a percentage of children who became ill, and the abstract does not report the absolute number of new diagnoses in each group. Readers comparing this figure with prevalence statistics should keep that distinction in mind.

The authors also split experimenters by timing. Those who had already sipped before baseline (n=1,236) formed Group A; those who first sipped during follow-up (n=423) formed Group B, each compared with its own matched controls. The subgroup findings below come from the pre-baseline group.

Sleep, bipolar disorder and self-harm stood out

Subgroup analyses in the pre-baseline exposure group pointed to specific outcomes rather than a uniform rise across all diagnoses. Pre-baseline alcohol exposure was associated with sleep-wake disorder (OR 1.39, 95% CI 1.06-1.82), bipolar disorder (OR 1.70, 95% CI 1.10-2.67), non-suicidal self-injury behavior (OR 1.91, 95% CI 1.20-3.08) and suicidal ideation (OR 1.51, 95% CI 1.22-2.15).

These are separate models for separate outcomes, and the paper does not present them as a ranked causal sequence. The bipolar estimate in particular rests on a wide interval, consistent with few observed cases in a two-year window among children this age.

The self-harm and suicidal-ideation results are the ones with the largest point estimates and intervals that stay clear of 1, which is why they carry weight in the authors' argument that children deserve attention in alcohol prevention.

Only rising sipping tracked with new diagnoses

Among the 423 children who began sipping during follow-up, LCTM identified two patterns over the two years: an increased-sipping class (Class 1, 56.7%) and a decreased-sipping class (Class 2, 43.3%). Both were compared with the same matched controls without alcohol use.

Only Class 1 was significantly associated with new mental disorders (OR 1.48, 95% CI 1.02-2.14). Class 2 was not (OR 0.82, 95% CI 0.52-1.26). This is the study's clearest internal comparison: the same exposure label, at least one sip, produced different risk depending on whether sipping was rising or falling across follow-up.

Suicidal ideation behaved differently. Both classes showed elevated odds relative to controls (Class 1: OR 2.38, 95% CI 1.22-4.71; Class 2: OR 2.64, 95% CI 1.29-5.44). So the trajectory split separated who developed a mental disorder overall, but not who reported suicidal ideation.

Sex-specific estimates in the Class 1 and Class 2 comparison were not statistically significant in either stratum, with intervals crossing 1 for both males and females.

Shared risk factors may explain part of the link

The authors do not claim that sipping causes mental disorders. They propose that early experimentation may act as a composite behavioral marker pointing to a set of shared genetic, temperamental and environmental risk factors. Children high in impulsivity or sensation-seeking, or with weaker executive functioning, may both start drinking earlier and be more likely to develop psychiatric problems.

They also name family environment as a candidate third factor: poor parental supervision, family conflict, parental substance use, chronic maltreatment and peer bullying could drive both outcomes. The introduction raises the reverse pathway too, the self-medication hypothesis, in which children with depression or anxiety initiate alcohol as coping. The design, which excludes children with a baseline diagnosis, reduces but does not eliminate that possibility.

This is the authors' interpretation of an association, not a tested mechanism. The paper reports no mediation or genetic analysis that would separate these pathways.

Minimal exposure and a short window limit the claim

Most participants reported only minimal exposure, often a single sip. That limits any dose-response reading: the authors state they could not see how odds rose as cumulative ethanol intake increased, and describe their ethanol analysis as preliminary. The exposure variable is therefore about experimentation, not about drinking quantity or frequency.

Alcohol use was self-reported by children, which the authors flag as a source of recall or reporting bias. Baseline differences between experimenters and non-drinkers in parental age, parental education, parental anxiety symptoms and family history were handled as covariates, but the authors acknowledge residual confounding could not be eliminated.

The two-year horizon also constrains what could be observed. The authors note that the young age of participants limited the number of cumulative cases, especially for disorders that typically peak in late adolescence or adulthood, and that alcohol effects on brain development may need five to six years to become apparent. The sample was drawn from an ABCD database primarily comprising individuals of White European descent, so the authors caution against generalizing to other ethnic populations.

Escalating sipping is the group to watch

The most decision-relevant result is not the overall 25% figure but the contrast between the two sipping trajectories. Children whose sipping increased over two years had higher odds of a new mental disorder (OR 1.48), while those whose sipping decreased did not (OR 0.82). A one-time report of "has sipped" would have merged these two groups and blurred the difference.

Suicidal ideation breaks that pattern, appearing elevated in both trajectory classes with overlapping intervals (2.38 and 2.64). For that outcome, the paper's evidence does not support using trajectory direction to triage risk.

Taken together, the study's own framing points to children under 12 as a priority population for alcohol prevention, and its internal comparison suggests that repeated assessment of whether sipping is rising, rather than a single yes/no question, is what separates the higher-risk group within that population.

Sources used in this answer