Why we need a broader view of adolescent mental health

Adolescence is a period of profound change. It is also a time when mental health difficulties can emerge or become more visible. These difficulties are often grouped into two broad domains: internalising problems, such as anxiety and depression, and externalising problems, such as attention-deficit hyperactivity disorder (ADHD) symptoms and conduct problems. We know that both genetic differences and environmental experiences contribute to these outcomes. The harder question is how they work together. Does an adverse environment amplify an adolescent’s genetic vulnerability? Or might a supportive or well-matched environment reduce it? These examples of gene-environment interaction are intuitively compelling but have proved difficult to demonstrate consistently. Many previous studies have tested one polygenic score – capturing a person’s genetic risk for a given trait – against one environmental exposure at a time. While this approach is valuable for focused questions, adolescent mental health is unlikely to arise from isolated influences. Genetic propensities overlap across traits, while family life, parenting, school experiences, and life events are also interconnected and unfold across development. We therefore wanted to take a broader view. Our first aim was to find out how well genetic propensities and environmental experiences predicted adolescents’ emotional and behavioural difficulties, both separately and when considered together. Our second aim was to test whether particular environments strengthened or weakened the links between genetic propensities and mental health. Finally, we wanted to examine whether adolescents’ genetic propensities were related to the environments they experienced, as this could help us understand any interactions we found. Figure 1 summarises what we measured, the questions we asked, and the study’s main findings.

Our approach

Our study drew on the Twins Early Development Study (TEDS), which has followed twins born in England and Wales since the mid-1990s. We analysed data from 3,337 unrelated 16-year-olds, combining a rich set of genetic, environmental, and mental health measures. Rather than relying on a single genetic indicator, we combined 13 polygenic scores indexing propensities related to psychiatric and neurodevelopmental traits. We also brought together multiple measures of the home, family, parenting, school, and wider environment collected when participants were aged 9, 12, and 16. The outcomes included adolescent- and parent-reported symptoms of externalising and internalising symptoms. This breadth created a methodological challenge: many predictors were related to one another, making it difficult to identify which ones are driving an association. We therefore used methods designed to consider many related factors at the same time while reducing the risk of chance findings. We tested whether genetic propensities and environmental experiences predicted adolescents’ emotional difficulties, such as anxiety and depression, and behavioural difficulties, such as ADHD symptoms and conduct problems. We considered genetic and environmental factors first separately and then together, before testing whether certain environments strengthened or weakened the links between genetic propensities and mental health. We also examined gene–environment correlation: whether genetic propensities were related to the environments adolescents experienced. This helped us assess whether any apparent interaction between genes and environments might partly reflect an existing link between the two.

What we found

Overall, both genetic and environmental factors predicted adolescent mental health, but their contributions were separate rather than combined. Across the outcomes, the polygenic scores accounted for an average of 3% of the differences between adolescents. This level of prediction is typical for current polygenic scores of complex mental health traits. It does not mean that genetics make only a small contribution to mental health; rather, today’s polygenic scores capture only a small part of the many genetic differences involved. The predictions were stronger for externalising than for internalising symptoms. Environmental measures accounted for an average of 7%, and models combining genetic and environmental information accounted for about 9% on average. Among the environmental measures, parental feelings—including frustration and impatience—and perceptions of household chaos were recurring predictors of emotional and behavioural symptoms. The timing and reporter also mattered. Parent-reported environments earlier in childhood were more predictive than parent-reported environments at age 16. Adolescents’ own reports of their environment at 16 were particularly informative for their self-reported symptoms. This may reflect adolescents’ growing autonomy and the increasing importance of their own perceptions. However, stronger associations may arise when the same person reports on both their environment and symptoms, particularly when both are assessed at the same time. We did identify gene-environment interactions for internalising symptoms, but not for externalising symptoms. On average, gene-environment interactions accounted for around 1% of variance. Some patterns suggested that adverse environments may amplify genetic vulnerability. For example, adolescents with higher genetic propensities for neuroticism or schizophrenia reported more anxiety or depression when they had experienced greater household chaos at age 9. At age 12, unpredictable or harsh environments were linked to more emotional difficulties among adolescents with higher genetic propensities for anxiety, depression or schizophrenia. At age 16, high levels of parental control and monitoring were linked to more anxiety or depression among adolescents with higher genetic propensities for anxiety, depression, schizophrenia or bipolar disorder. Other patterns were less straightforward. In some cases, the link between genetic propensity and mental health changed direction depending on the environment. One possible explanation is person–environment fit: some people may cope better in environments that suit their individual characteristics, while a mismatch between a person’s characteristics and the demands of their environment may contribute to difficulties. For example, a highly structured environment might be more challenging for an adolescent with a greater propensity towards ADHD, whereas a less structured setting might suit them better. However, these findings should not be read as simple prescriptions about parenting or home structure. The effects were small, some were complex, and replication is essential.

Why small effects still matter—and why caution matters more

An absent or small interaction is not a disappointing result. It tells us something important about the architecture of adolescent psychopathology. In our data, adding genetic information improved prediction based on measured environments, but the two sources of information were generally distinct rather than interacting in large or consistent ways. This is especially important because claims about genes and environments can easily become deterministic. Polygenic scores indicate genetic probabilities, not diagnoses. Likewise, links between environmental factors and symptoms do not prove that the environment directly cause the mental health difficulties. These associations may arise for several reasons. Children can influence the environments they experience, while parents pass on genes and also shape the home environment. For example, a parent with a greater tendency towards anxiety might monitor their child more closely while also passing on genetic propensities related to anxiety. Other unmeasured family factors could also affect both the home environment and adolescent mental health. As our earlier analyses suggested some overlap between the genetic and environmental predictions for anxiety and depression, it is difficult to separate genuine gene–environment interactions from these related processes. This reinforces the need for careful interpretation.

Where the research goes next

The study also highlights some of the field’s current limitations. Today’s polygenic scores capture only a small fraction of genetic differences, especially for internalising symptoms in young people. The sample included participants of European ancestry because available polygenic scores perform less accurately across other ancestry groups, limiting generalisability. Some interaction analyses also had smaller samples than the main analyses, and the environmental measures focused heavily on the home and parenting context. Future work will need larger and more diverse samples, stronger genetic discovery studies of childhood and adolescent traits, richer measures of schools and broader social contexts, and designs that can better distinguish environmental effects from shared family or genetic factors. Replicating the findings in other studies and samples will be crucial before any specific gene-environment interaction can inform interventions. The main lesson is that adolescent mental health reflects many small, overlapping influences. Considering genes and environments together bring us closer to this complexity, but it also cautions us against simple explanations. Doing so responsibly means considering how strong the associations are, how certain we can be, and the limits of what the data can tell us.