Skip to main content
Search
Main content
BMJ mental health
Published

Genetic investigation of non-affective psychosis and depression as causal risk factors for dementia

Authors

Valentina Escott-Price, Emily Simmonds, Michael J Owen, Michael O'Donovan

Abstract

BMJ Ment Health. 2026 Jun 24;29(1):e302424. doi: 10.1136/bmjment-2025-302424.

ABSTRACT

BACKGROUND: Major psychiatric disorders are associated with increased risk of dementia but establishing whether psychiatric disorders causally increase dementia risk is challenging because dementia pathology can precede clinical diagnosis by decades. Prodromal psychiatric symptoms may arise long before cognitive decline, leaving open the possibility of reverse causation.

OBJECTIVE: We aimed to determine whether non-affective psychosis or depression credibly causally influence dementia risk using a design robust to reverse causation. We tested whether people with psychiatric disorders who later develop dementia show reduced genetic liability to Alzheimer's disease (AD) compared with dementia cases without such history.

METHODS: We compared AD genetic liability, measured by polygenic risk scores (PRS) among dementia cases (N=7936) with and without prior non-affective psychosis (N=56) or depression (N=937) in the UK Biobank. We examined whether schizophrenia or major depressive disorder (MDD) PRS correlates with dementia liability to assess whether shared trait liability contributes to the association.

FINDINGS: Dementia cases with prior non-affective psychosis or depression had lower AD genetic liability than those without a psychiatric history (psychosis: B=-0.29, 95% CI (-0.54 to -0.05), p=0.036; depression: B=-0.12, 95% CI (-0.18 to -0.05), p=0.0004), which is inconsistent with the hypothesis that the association between psychiatric disorders in dementia is explained by prodromal dementia effects. After excluding people with psychiatric diagnoses, neither schizophrenia nor MDD liability showed the negative correlations with AD liability in people with dementia expected if trait liability to those disorders per se contributed to dementia risk. Our findings instead are consistent with exposure to the disorders themselves as being associated with dementia.

CONCLUSIONS: Our findings are consistent with the hypothesis that psychiatric disorders are associated with increased vulnerability to dementia.

CLINICAL IMPLICATIONS: Identification of potentially modifiable mechanisms for the association and optimal management of non-affective psychosis and depression may help reduce long-term dementia risk and inform prevention strategies.

PMID:42342369 | DOI:10.1136/bmjment-2025-302424

UK DRI Authors

Prof Valentina Escott-Price

Group Leader & Research Division Lead

Using Big Data, machine learning and AI to accelerate discoveries into dementia

Prof Valentina Escott-Price