Shedding light on weight and height related health matters using routinely-collected electronic health records

Overview

Three million people die each year from obesity-related diseases. A research team from Population Data Science Swansea funded by HDR UK and led by Dr. Fatemeh Torabi and Prof. Ashley Akbari have created a national data asset including body mass index (BMI) measures for the entire population of Wales. They achieved this by linking data from General Practice (GP), hospital, maternity and school health check records and unifying them into a single dataset. This research ready data asset (RRDA) will help to answer critical questions about the population’s health.

The challenge

Accurate, long-term data on BMI is an important part of nearly all health data research. It helps researchers understand shifting patterns in population’s health, assess new ways to treat or prevent obesity-related diseases, and forecast people’s disease risks. Although these measures can be sourced regularly through a range of interactions with health services, much of the population’s BMI data is often not recorded during routine health checks, leaving researchers with a lot of missing BMI data. In 2022, coverage was just 34% for adults and 3% for children and young people registered with a general practice (GP) in Wales.

The solution

To build a fuller picture of BMI for the Welsh population, Dr. Michael Jeanne Childs, the lead author, linked data from GP records, hospital admissions, maternity records, and community health checks.

This comprehensive data linkage was conducted within the Secure Anonymised Information Linkage (SAIL) Databank, the national Trusted Research Environment (TRE) for Wales, which covers the whole population of Wales. The team developed a methodology to standardise these data sources, unifying the results into an RRDA for BMI data spanning 5.1 million people over 23 years, from January 2000 to December 2022 at the time of the output creation.

Dr Childs explains: “There are quite a lot of inconsistencies with the electronic health record (EHR) data sources that we have access to for BMI; some are recorded as height and weight, some as centimetres and some in inches, for example. So the methodology we’ve created addresses these inconsistencies to have harmonised results.”

The impact

The team have created a reusable population-scale RRDA resource and made it open access so that other researchers can accurately repeat what they’ve done in other projects. This effectively establishes a longitudinal RRDA on BMI for use in research, representative of the Welsh population.

It has already been put to good use in many other research projects, enabling them to generate insights across different population subgroups stratified by BMI. The methodology can be used within SAIL and is transferable to other TREs across the UK that hold similar data sources that capture BMI or its component parts. In practice, this methodology is designed for researchers examining population-wide health outcomes using routinely-collected electronic health records at scale.

A particular strength of the RRDA is its accurate, long-term childhood obesity data; until now, childhood BMI data has been either self-reported or a one-off measurement taken in school. Coverage of children’s BMI data in Wales has increased by 44% as a result of this work.

“This resource captures BMI measurements over time,” explains Dr Torabi. “for instance, when a child visits the GP and their height and weight are recorded, the BMI methodology identifies that entry and incorporates it into what we term a BMI spine for the whole population of Wales. In practice, for anyone with an existing BMI-related record in one of the core data sources, the methodology captures, harmonises and integrates their records into the BMI spine, allowing us to build a far more complete and accurate picture of BMI patterns across the whole population of Wales.”

Looking ahead, Prof. Akbari and team hope the RRDA and associated methodology can continue to be used by others in SAIL and expanded across other TREs across the UK, adding value to research and leading to improvements and benefits for people’s lives and services.

What the impact committee said:

The impact committee praised this ‘broad and generalisable population-based study’ which used robust methodology to create unified BMI records across multiple sources.