A major peer-reviewed study published in Genomic Psychiatry has found that genetic risk for psychiatric disorders is far less specific to individual conditions than previously assumed, with substantial variation across diagnoses and clinical features.
Led by Dr Kenneth S Kendler, VIPBG distinguished professor of psychiatry at Virginia Commonwealth University, the research analysed data from more than two million individuals born in Sweden between 1950–1995. Using national patient and primary care registers, researchers assessed how much genetic liability linked to one disorder predisposes specifically to that condition vs others.
The team examined nine disorders: schizophrenia, bipolar disorder, alcohol-use disorder, ADHD, autism spectrum disorder, posttraumatic stress disorder (PTSD), major depression, anxiety disorder, and drug-use disorder.
Family genetic risk scores were calculated based on morbidity patterns in first- to fifth-degree relatives, with adjustments for cohabitation effects. Linear regression was then used to estimate the proportion of total genetic risk attributable to the index diagnosis, defined as genetic specificity.
Schizophrenia showed the highest genetic specificity at 73.1% (95% CI: 66.3%–79.8%), meaning nearly three-quarters of aggregate genetic risk in affected individuals related uniquely to schizophrenia. Bipolar disorder followed at 54.8%, and alcohol-use disorder at 54.1%.
A middle tier included ADHD at 48.2%, autism spectrum disorder at 47.5%, and PTSD at 47.4%.
Lower specificity was observed for major depression at 41.1%, anxiety disorder at 38.6%, and drug-use disorder at 29.5%. For drug-use disorder, less than a third of genetic risk was disorder-specific, with the remainder overlapping other conditions.
“What surprised us was the sheer range,” said Dr Kendler. “Schizophrenia carries a genetic signature that is overwhelmingly its own. Drug use disorder, by contrast, looks more like a downstream expression of genetic risks that cut across many conditions. That difference has real implications for how we design genetic studies and how we think about diagnostic categories.”
Genetic specificity varied significantly by clinical characteristics. Greater recurrence was associated with higher specificity across all nine disorders, particularly bipolar disorder and ADHD.
For bipolar disorder, early-onset cases showed substantially higher specificity than late-onset cases. Hospital-treated patients had markedly higher specificity at 63% (95% CI: 60%–67%) compared with 31% (95% CI: 16%–46%) among those treated only in primary care, a difference exceeding 30% points (p < 0.001).
PTSD showed the opposite treatment-setting pattern, with higher specificity in primary care cases at 53% (95% CI: 50%–56%) than in hospitalised cases at 41% (95% CI: 37%–45%).
In major depression, hospitalised cases exhibited lower specificity, potentially reflecting impulsive behaviours, suicidal ideation, and substance-related crises associated with externalising risks such as ADHD, alcohol-use disorder, and drug-use disorder. Primary care depression may represent a more mood-specific genetic signal.
“Genetic specificity is not some abstract property locked inside the genome,” Dr Kendler said. “It moves. It responds to clinical features that every psychiatrist can observe at the bedside. A hospitalized bipolar patient and one seen only in primary care carry substantially different levels of genetic specificity.”
The findings align with molecular research, including a 2026 study published in Nature identifying a general psychopathology P-factor and subfactors. Internalising conditions such as major depression, anxiety disorder, and PTSD shared more than 90% of variance with the P-factor, while schizophrenia and bipolar disorder shared 35%, reflecting the hierarchy observed in this study.
Robustness checks supported the findings. Adjusting for comorbidity had minimal impact; for example, depression specificity shifted from 41.1% to 41.8% after excluding bipolar overlap cases. Sex differences were most evident in substance-use disorders, with men showing higher specificity (p < 0.001). Leave-one-out analyses confirmed expected dependencies between highly correlated conditions such as major depression and anxiety disorder.
The study relied on registry diagnoses rather than structured interviews, although Swedish registry data are well validated. The largely homogeneous Swedish population may limit generalisability. Family genetic risk scores differ from DNA-based polygenic scores, though previous research indicates consistency.
The study, titled The specificity of genetic risk for psychiatric and substance use disorders: Its modification by age at onset, recurrence, and site of treatment, was published on 3 March 2026 in Genomic Psychiatry.
“We have been debating whether psychiatric disorders are truly distinct since the 1800s,” Dr Kendler said. “Now we can actually put numbers on it. Some of our diagnostic categories carve nature much more cleanly at the genetic joints than others, and clinicians and researchers alike need to reckon with that.”