Design: population-based cohort study using linked nationwide Danish registries, with simulation against a random-mating counterfactual
Population: 4,244,585 Danish residents born 1969–2017, including 549,541 parental couples and 822,209 offspring
Primary outcome: parental correlation for psychiatric and somatic disorders, and its effect on prevalence and heritability
Effect: correlation 0.28 for any psychiatric disorder, 0.38 for schizophrenia, 0.11 for anorexia; simulated inflation of prevalence 1.8% and family-based heritability 12.5%

Heritability estimates for psychiatric disorders are built on the assumption that parents pair off more or less at random with respect to the trait. A Danish registry study of 4,244,585 residents born between 1969 and 2017 tested that assumption against real couples — 549,541 parental pairs with 822,209 offspring.

Parents are markedly alike. The correlation between partners for any psychiatric disorder was 0.28. Within specific disorders it was highest for schizophrenia at 0.38 and lowest for anorexia at 0.11, with cross-disorder correlations running from 0.01 to 0.26 — a parent with one diagnosis is more likely to have partnered with someone carrying a different one. Correlations for somatic illness were much weaker (0.03 to 0.14) and were almost entirely explained by psychiatric comorbidity. Simulation against a random-mating scenario suggested this pattern inflates the population prevalence of any psychiatric disorder by 1.8% and family-based heritability estimates by 12.5%.

The clinical translation is about family history rather than genetics. When a patient reports psychiatric illness in one parent, the prior probability of illness in the other is meaningfully higher than the population rate, and risk in that family concentrates rather than averages out. It also means a family-history-based heritability figure quoted to a patient is an overestimate of what the genes alone contribute.

Ask about both parents explicitly; a positive history on one side raises the odds on the other.
Ask across diagnoses, not only the one in front of you — cross-disorder partner similarity is substantial.
Expect risk to cluster within particular families rather than spread evenly.
Be careful quoting heritability percentages to families; the familial figure includes a partner-similarity component.
Where one parent is under your care, consider whether the other and the children have ever been assessed.

Why it matters

It undercuts the assumption behind every family-based heritability figure we quote, and it changes what a one-parent family history implies.

The statistics, in plain English

A partner correlation of 0.28 is large for a population measure — for comparison, the somatic-illness correlations were 0.03 to 0.14, and those largely vanished once psychiatric comorbidity was accounted for. The 1.8% and 12.5% inflation figures come from simulating what the population would look like under random mating, so they are model outputs rather than observed quantities; the observed part is the correlation itself, which is measured on more than half a million couples and has a standard error of 0.003.