Epidemiological Studies of Adiposity and Health Outcomes in Later Life
Bowman, Kirsty Holly
Thesis or dissertation
University of Exeter
Reason for embargo
Would like to publish one of the chapters presented in this thesis,
Globally, there has been a substantial rise in the prevalence of obesity amongst older persons (≥65 years) since the 1980s. Conflicting evidence exists on the impact of this trend for health outcomes. A 2013 meta-analysis documented no statistical difference for mortality between those within the body mass index (BMI) defined Obese-1 (30.0-34.9 kg/m2) range and those within the conventional BMI Normal (18.5-24.9) range, sparking vigorous debate. The reduced or similar mortality risks for those within the BMI Obese-1 range relative to those within the BMI Normal range has been termed the “obesity paradox”. Clarifying associations of obesity with health outcomes could have implications for intervention in later life. I aimed to examine this paradox by assessing the length of follow-up, the BMI referent group, the inclusion of smokers plus those with conditions associated with weight loss, and alternative measures of adiposity. I analysed >955,000 electronic health records from the UK Clinical Practice Research Datalink for patients aged ≥60 years. I showed reduced mortality risks for those within the BMI Obese-1 range relative to those within the BMI Normal range across each age group. Mortality risks were reversed after restricting the analysis to ‘healthier agers’, demonstrating that the paradox is partly explained by the inclusion of smokers, adults with conditions associated with weight loss, and the chosen BMI referent group. Similarly, I document no support for reduced dementia risks for those within the BMI Obese-1 range. Additionally, there was an increased risk for incident of coronary heart disease and diabetes for those within the BMI Obese-1 range. BMI does not capture body composition changes with ageing. I found additional adiposity measures improved the mortality prediction compared to BMI only using the UK Biobank comprising >200,000 older volunteers. Mortality risks were increased for those who were centrally obese across the BMI Normal to Obese-1 range. In conclusion, I have shown the heterogeneity of older adults can result in disparate risk estimates for the association between BMI and health outcomes. I provide additional evidence that reliance on BMI measures only may miss those at increased risk for health outcomes due to central adiposity. My results provide no support, in relatively healthy older adults, for the hypothesised obesity paradox in later life.
PhD in Medical Studies