Assessing Representativeness and Bias in Large Prospective Biobanks: Insights from Our Future Health

Speaker: Melinda Mills, MBE FBA FAcSS, Leverhulme Centre for Demographic Science and Nuffield Professor of Demography. 

Bio: Melinda Mills is a Professor of Demography and Population Health, Director of the Leverhulme Centre for Demographic Science and Demographic Science Unit at NDPH and Nuffield College. Her work focuses on demographic change, combining multiple types of high-dimensional data and advanced statistical methods. In addition to others, she has held both an ERC Consolidator and ERC Advanced Grant, examining the intersection of social and genetic factors. She has served on No 10's Data Science Advisory Group, as an advisor on SAGE (SPI-B), and as one of three special Advisors to the European Commissioner of the Economy. She is also a Trustee of the UK Biobank and on the Scientific Advisory Boards of Our Future Health, LifeLines, Health & Retirement Survey. She received an MBE in 2018 for her research contributions and an Honorary Doctorate in 2025 from the EUI for her work in sociogenomics. She also holds a part-time position at the Department of Economic, Econometrics & Finance, University of Groningen and Department of Genetics, University Medical Centre Groningen, Netherlands.

Abstract:  The scale and coverage of biobanks offer unique opportunities to advance multiple research domains. This talk focuses on Our Future Health (OFH), which is a new prospective study aiming to recruit 5 million UK-resident adults. It examines the latest December 2025 data release where baseline phenotypic data are available for >1.9 million participants. It assesses representativeness across multiple domains of OFH against national estimates and other cohorts, particularly the UK Biobank (UKB). Phenotypes that are assessed include: self-reported health-related behaviours, geolocation, disease diagnoses and medication, in- and outpatient visits, cancer registry and cause of death. Comparisons are made across sociodemographic, lifestyle and health-related characteristics and case prevalence concordance and known clinical correlates are compared across OFH and UKB.  Medication-use patterns and cancer prevalence are also examined across age-related gradients. The talk concludes with a reflection on the consequences of participation bias for different research questions and types of weighting.

There will be tea/coffee and cake 30 min prior to the Lecture.

Join the meeting online