Summary
This paper demonstrates how collider bias—a phenomenon whereby selection into a non-representative sample induces spurious associations between independent variables—undermines causal inference in observational COVID-19 studies. Using UK Biobank data, the authors show that participants tested for SARS-CoV-2 were highly selected on multiple traits (genetic, behavioural, cardiovascular, demographic, anthropometric), distorting observed risk factor associations. The work argues that appropriate sampling strategies at the study design stage are essential to mitigate collider bias, rather than attempting to correct for it post-hoc.
UK applicability
This finding is directly applicable to UK research institutions and public health authorities interpreting COVID-19 epidemiological studies. UK Biobank itself and NHS-linked observational studies should be reassessed for collider bias when identifying COVID-19 risk factors, with implications for evidence-based public health guidance and health communications in the United Kingdom.
Key measures
Selection patterns for genetic, behavioural, cardiovascular, demographic, and anthropometric traits in UK Biobank participants tested for COVID-19; associations between variables induced by sampling bias
Outcomes reported
The study analysed UK Biobank data to demonstrate how collider bias distorts associations between risk factors for COVID-19 infection and disease outcomes in non-representative samples. It identified the mechanisms by which selection into testing or hospitalisation induces spurious associations and discusses mitigation strategies.
Topic tags
Dig deeper with Pulse AI.
Pulse AI has read the whole catalogue. Ask about this record, its theme, or how the findings apply to UK farming and policy — every answer cites the underlying studies.