Media release

From: Springer Nature

Reassessing associations between loneliness and disease

Loneliness may not be a direct cause of disease, a genetic study of over 400,000 participants, published in Nature Human Behaviour, suggests. However, the findings indicate that managing risk factors related to loneliness could still improve long-term health outcomes.

Loneliness is the subjective feeling of social disconnection and has been linked to numerous health issues, such as depression, diabetes, and cardiovascular disease. Understanding this link is important for developing effective interventions, and although previous studies have documented some associations, for most diseases, these associations and potential causal relationships have not been established.

Jihui Zhang and colleagues analysed data of 476,100 participants from the UK Biobank over a median follow-up of 12.2 years. Observational analyses examined the association between loneliness and 56 diseases across 14 categories. Observational results showed that loneliness is likely associated with an increased risk in 13 of 14 disease categories and 30 of 56 diseases. The diseases most strongly associated with loneliness were posttraumatic stress disorder, depression, anxiety, schizophrenia, and chronic obstructive pulmonary disease. Of the 30 diseases associated with loneliness, 26 had genetic data available for further analyses. The results of these genetic analyses indicated non-causal associations with loneliness for 20 out of 26 diseases, including cardiovascular diseases, type 2 diabetes, and chronic liver disease. The authors suggest this may indicate that loneliness is a surrogate marker, which may predict these diseases in patients rather than being a direct cause of them.

These findings highlight the importance of addressing risk factors related to loneliness, such as unhealthy behaviors and comorbidities, to improve health outcomes. The authors add that future research is needed to explore the mechanisms that link loneliness to disease, and should include more diverse participant samples.