Body mass index and cause-specific mortality - population-based cohort study using record linkage

Study type
Protocol
Date of Approval
Study reference ID
16_174
Lay Summary

Previous studies have shown that body mass index (BMI), a measure that captures a person's weight taking into account their height, is associated with the overall risk of dying at a given age. However, few large studies have explored how BMI is related to dying from specific causes. Our objective is to comprehensively investigate how BMI is associated with broad and specific causes of death. We will used primary care data (which includes information on a person's BMI) linked to cause of death data as recorded on death certificates. We will fit statistical models to look at how BMI is associated with each of a wide range of causes of death, taking into account information on other factors like smoking that could be linked to both BMI and the risk of dying from particular diseases. We will look at whether the role of BMI is different for different subgroups of the population. Finally, within groups of people with similar BMI levels, we will estimate the proportion dying from different causes at specific ages.

Technical Summary

Body mass index (BMI) is associated with all-cause mortality, but few large studies have explored associations with death from specific causes. Our objective is to comprehensively investigate associations between BMI and specific causes of death. We will used CPRD primary care data linked to Office of National Statistics mortality data. All individuals with a BMI record in the Clinical Practice Research Datalink (CPRD) will be included. Outcomes will be underlying causes of death, using categorisations developed as part of the Global Burden of Disease project. We will look at both broad and specific categories/groupings of causes of death. We will use Cox regression models based on cause-specific hazards to model the associations between BMI and each cause-specific mortality outcome, adjusting for key potential confounders. We will use cubic splines to allow for non-linearity, and we will fit interactions to investigate effect modification by individual level factors. We will also use competing risks methods to estimate cumulative incidences for each outcome, stratified by BMI category.

Health Outcomes to be Measured

All-cause mortality Cause specific mortality (Global Burden of Diseases classification)

Collaborators

Krishnan Bhaskaran - Chief Investigator - London School of Hygiene & Tropical Medicine ( LSHTM )
David Leon - Collaborator - London School of Hygiene & Tropical Medicine ( LSHTM )
Ian Douglas - Collaborator - London School of Hygiene & Tropical Medicine ( LSHTM )
Isabel dos-Santos-Silva - Collaborator - London School of Hygiene & Tropical Medicine ( LSHTM )
Liam Smeeth - Collaborator - London School of Hygiene & Tropical Medicine ( LSHTM )

Linkages

ONS Death Registration Data;Patient Level Index of Multiple Deprivation