IJPDS International Journal of Population Data Science

Researchers from Queen Mary University of London, Swansea University, and the University of Edinburgh have successfully tested the theory of using health records to provide valuable and up-to-date information about the population in a local area. They used anonymised electronic health records from north-east London to calculate the population of the region to give an overall count, as well as analysing the population by age, sex, and local authority. In addition, they were also able to calculate the number and size of households using a novel method that anonymously links people who share the same address in their patient record.

Statistics on the number of people and where they live are used by local governments to plan infrastructure, schools, housing and health services, and to produce regional insights on health, the economy and education. These statistics currently come from the National Census, which is a survey of all households in England and Wales. However, the Census is only conducted every 10 years, so population estimates between censuses get progressively less accurate. However, almost everyone is registered with their local GP, so anonymised health records have the potential to provide valuable and up-to-date information about the population in a local area.

The research team compared the results to the corresponding published Office for National Statistics Census 2021 estimates and found that health record estimates are close enough to those of the Census to provide reliable, real-time population estimates for local applications in the interim between Censuses.

The study, published in the International Journal of Population Data Science (IJPDS), shows the population estimate from health records was 7% higher than that of the Census. The number of households estimated through health records was 9% lower than the Census, with agreement varying by size of household (poorer for households of eight people or more) and by borough (poorer for the London Borough of Hackney).

The researchers also compared their estimates to others produced from multiple data sources by ONS and found better agreement with these sources than with Census data.  While differences between Census and health record counts could be because some people left the area without deregistering from their GP, this was not found to be the case. This highlights the complexities of counting people in households using administrative data, especially in neighbourhoods where people move house frequently.   

Lead author Marta Wilk said “The study demonstrates that pseudonymised data from health records can be used to produce up-to-date population statistics, comparable to those from the Census and more complex linked datasets. These estimates can support planning, especially in regions that are predicted to have marked population growth over the coming decade."

 

Click here to read the full article

Marta Wilk, Wolfson Institute of Population Health, Queen Mary University of London, UK

Wilk, M., Harper, G., Firman, N., Dibben , C., Fry , R. and Dezateux , C. (2025) “Estimating households and populations from primary care electronic health records: comparison with Office for National Statistics Census 2021 aggregated estimates”, International Journal of Population Data Science, 10(1). doi: 10.23889/ijpds.v10i1.2958.