When Parents Are Hospitalized for Mental Illness, Family Genes May Drive Children’s Risks

One of the largest studies ever conducted on the intergenerational echoes of mental illness has delivered a finding that could reshape how scientists and policymakers think about risk. Children whose parents are hospitalized for severe psychiatric disorders do face markedly higher rates of suicidal behavior, violent offending, and violent victimization in adolescence and early adulthood. But according to a sweeping analysis of nearly three million Swedes, much of that elevated risk appears to stem not from the hospitalization itself, nor even from the direct consequences of a parent’s illness, but from vulnerabilities that run through the family tree.

The research, published in SSM – Population Health, drew on Sweden’s extraordinary national registries to follow 2,978,645 individuals born between 1975 and 2005. The researchers, led by Gloria Yuxuan Gu and colleagues including Antti Latvala, Henrik Larsson, and Paul Lichtenstein, linked records spanning births, hospitalizations, criminal convictions, and deaths. Roughly 6.4 percent of children under 15 had experienced a parent being hospitalized for a severe mental illness such as schizophrenia or bipolar disorder, an affective disorder such as depression or anxiety, or a substance use disorder. About one in ten children in Sweden experiences parental psychiatric illness during childhood, a figure comparable to other high-income countries.

At first glance, the raw numbers are stark. Among young people whose parents had never been psychiatrically hospitalized, roughly 2 percent experienced one of the violence-related outcomes by early adulthood. Among those exposed to parental hospitalization for severe mental illness, the figure rose to nearly 4 percent for violent victimization. For children whose parents were hospitalized for substance use disorders, 9.17 percent went on to be convicted of a violent crime. In absolute terms, the researchers calculated an excess of roughly 17 to 66 additional cases per 1,000 exposed offspring, with the heaviest burden attached to violent offending after parental substance-related hospitalization.

Statistical models confirmed the pattern. In conventional population-level analyses that adjusted for birth year, sex, county of residence, and parental education and age, children of psychiatrically hospitalized parents showed hazard ratios ranging from 1.56 to 2.76, meaning a 56 to 176 percent increase in the hazard of a violent outcome. Hospitalization for accidents, by contrast, carried lower risks, between 1.39 and 1.51. These are the kinds of estimates that have populated the literature for years, and they have fueled calls for screening and preventive outreach aimed at children of mentally ill parents.

But the Swedish team went a crucial step further. Using a cousin-comparison design, they compared children whose parents had been hospitalized with their own cousins whose parents had not, holding constant the extended family environment and, critically, a substantial share of shared genetics. Cousins share grandparents and inherit, on average, about 12.5 percent of their genes in common, so differences between cousins cannot easily be explained by the familial liabilities that pervade whole families. When the analysis moved from between-family to within-family comparisons, the hazard ratios dropped substantially, falling to a range of 1.27 to 1.65. On average, the associations shrank by 22.4 percent for severe mental illness, 25.4 percent for affective disorders, and 34.5 percent for substance use disorders.

The accident comparator sharpened the picture further. Accidental hospitalizations share many of the disruptive features of psychiatric admissions, including a parent’s temporary absence, financial strain, and upheaval in daily routines, but they are not systematically tied to the familial psychiatric liability that cousin comparisons are designed to control. Notably, the attenuation from population-level to within-family estimates was smaller for accident hospitalizations, around 15.5 percent on average, than for psychiatric hospitalizations. This asymmetry suggests that the excess risk following psychiatric admissions is driven primarily by disorder-specific familial transmission, including shared genetic liability, related behavioral traits such as impulsivity and emotional dysregulation, and the chronic family environments that travel with these conditions, rather than by the acute disruption of a parent’s absence alone.

The study also probed whether timing matters. Developmental science has long proposed that early childhood represents a sensitive period for attachment and emotional regulation, and that adolescence is critical for social and behavioral development. Yet when the researchers stratified exposure by the child’s age, dividing childhood into windows of 0 to 5, 6 to 10, and 11 to 14 years, the associations were largely stable across periods. An apparent stronger effect of early exposure to parental substance-related hospitalization vanished entirely once familial confounding was accounted for. The authors argue that what looks like a biologically privileged early window in population data may simply reflect the fact that earlier and longer exposure accumulates more socioeconomic disadvantage. Their practical implication is pointed: preventive services should be maintained continuously across childhood rather than concentrated at any single developmental stage.

Dose-response analyses told a similar story. At the population level, each additional cumulative month of parental hospitalization was associated with higher offspring risk, and accidental hospitalizations showed a particularly clean 6 percent increase in risk per month, a pattern that survived within-family comparison. For psychiatric exposures, however, the cumulative-duration signal largely dissolved after adjusting for familial confounding, with the exception of substance use disorders, whose associations remained significant but substantially attenuated. Together with the accident findings, this hints at an intriguing possibility: the disruption of caregiving itself, as captured by accident-related admissions, may have a more direct causal role than previously appreciated, and deserves investigation in its own right.

Perhaps the most actionable results concern what happens next in young people’s lives. When the researchers additionally adjusted for offspring educational attainment, employment status, and civil status one year before the outcome, the hazard ratios fell further, to between 1.20 and 1.48 for psychiatric exposures. This indicates that life-course transitions in emerging adulthood, finishing school, finding work, forming stable relationships, partially channel the residual risk. These are precisely the domains that social policy can reach. The authors suggest that existing programs in education and labor market integration could be evaluated with intergenerational risk transmission as a secondary outcome, turning an etiological insight into a set of modifiable levers.

The findings come with caveats. Cousin comparisons control only part of shared genetic liability, so the within-family estimates are best read as upper bounds of any true environmental effect. Mediation analyses assume, but cannot prove, causal pathways. And Sweden’s generous welfare state buffers many of the socioeconomic consequences of parental illness, meaning the environmental component could be larger in countries with weaker safety nets. Still, the central message is hard to escape: the alarming hazard ratios reported in earlier registry studies should not be interpreted as the causal impact of a parent’s hospitalization, and family-focused prevention calibrated to those raw estimates is likely to yield smaller returns than hoped. The better strategy, the authors conclude, is twofold, sustained support for families carrying high underlying vulnerability, and deliberate protection of the educational, occupational, and relational transitions through which whatever residual environmental risk remains is ultimately expressed.

Subject of Research: Intergenerational transmission of risk for suicidal behavior, violent offending, and violent victimization following parental psychiatric hospitalization

Article Title: Childhood Parental Psychiatric Hospitalization and Offspring Suicidal Behavior, Violent Crime, and Violent Victimization: A Swedish Population-Based Cousin Comparison Study

Article References: Gu, G. Y., Latvala, A., Kuja-Halkola, R., Zhou, M., Larsson, H., D’Onofrio, B. M., Chang, Z., & Lichtenstein, P. (2026). Childhood Parental Psychiatric Hospitalization and Offspring Suicidal Behavior, Violent Crime, and Violent Victimization: A Swedish Population-Based Cousin Comparison Study. SSM – Population Health, Article 101978. https://doi.org/10.1016/j.ssmph.2026.101978

Image Credits: AI Generated

DOI: 10.1016/j.ssmph.2026.101978

Keywords: parental psychiatric hospitalization, suicidal behavior, violent crime, violent victimization, cousin comparison, genetic confounding, Swedish registries, substance use disorders, severe mental illness, life-course transitions, causal inference, public health