Monash University

Restricted Access

Reason: Access restricted by the author. A copy can be requested for private research and study by contacting your institution's library service. This copy cannot be republished

Obesity, diabetes and disability pathways, predicting risk and projections

posted on 2017-02-22, 04:15 authored by Wong, Evelyn Ee Ling
Background and objectives In this thesis I set out to investigate the impact of current obesity and diabetes trends on physical disability against a backdrop of an ageing population and increased obesity and diabetes prevalence over the last three decades. Between 1980 and 2000, obesity and diabetes prevalence in Australia more than doubled. If current trends continue, it is estimated that by year 2025, over 1 in 3 adult Australians will be obese and over 1 in 10 will have diabetes. My hypothesis at the beginning of my PhD was that the increases in obesity and diabetes prevalence over time would be associated with increases in disability prevalence, thereby impacting on individuals, their carers and society. In this thesis I aimed to delineate the challenges to healthy ageing from increased obesity and diabetes prevalence and to develop the building blocks toward estimating risk of developing disability in old age given individual risk profiles in mid-life. To better understand the challenges our ageing population face required in-depth understanding of the association between obesity, diabetes and the development of disability. To this end, I investigated and contributed to the evidence in three broad areas: 1. Quantification of the associations between obesity, diabetes and disability 2. Identification of the key modifiable mid-life predictors of disability – including the development of a risk algorithm in mid-life to predict disability in old age 3. Estimation of the preventable burden of disability attributable to obesity and diabetes Analyses To achieve the objectives of my PhD, I undertook six projects utilising various epidemiological and statistical methods, which included systematic review and meta-analyses, binomial logistic regression, multinomial logistic regression, Cox proportional hazards regression, predictive analysis and life table methods. Individual level data from large cohort studies were used for all analyses except the systematic review and meta-analyses. The cohort studies used were the Australian Diabetes, Obesity and Lifestyle Study (AusDiab), the Melbourne Collaborative Cohort Study (MCCS) and the Framingham Offspring Study (FOS), an American cohort study. Two systematic reviews with meta-analyses were conducted to pool risk estimates for the association between (1) diabetes and disability and (2) obesity and disability. Following the demonstration of associations between both diabetes and obesity with disability, a subsequent project was undertaken to identify other key modifiable risk factors in mid-life that were significantly associated with disability in old age. Prior to this risk algorithm, there was no way of quantifying risk of an overall health outcome, that of disability, from a combination of risk factors in mid-life. These mid-life risk factors were then combined to develop a risk prediction algorithm for disability. In both the development of this disability risk prediction algorithm and in the systematic review we demonstrated that obesity, measured using BMI, was a significant predictor of disability. Therefore, we subsequently undertook two other projects to analyse the effects of obesity on disability in more detail. First, we investigated which adiposity measure is most predictive of disability by comparing the predictive value of BMI, waist circumference and a range of body composition measures, such as fat mass and percentage fat, on disability. Secondly, we investigated the extent to which duration of obesity affects the obesity-disability relationship over and above BMI attained. Following the development of a disability risk prediction algorithm using potentially modifiable risk factors in mid-life for disability in old age, we undertook a project applying such a disability risk prediction algorithm on simulated populations to estimate the effects of changes in obesity and diabetes trends on disability prevalence in Australia. Key findings The two systematic review and meta-analyses on diabetes and disability and obesity and disability demonstrated that: 1. Having diabetes compared to not having diabetes increases the risk of disability by 50-80% across all types of physical disability. 2. Obesity and disability showed a positive graded association, such that increasing severity of adiposity from overweight through to the obesity class 1, and class 2 and above increased the magnitude of association with limitations to activities of daily living (ADL). We identified smoking, diabetes and obesity as significant modifiable risk factors in mid-life for disability in old age after adjusting for age and sex. We developed a risk prediction algorithm with those factors including age and sex to predict likelihoods of disability, death and of surviving free of disability over a 13-year period. We demonstrated combined effects of these risk factors. For example, we demonstrated that a 45 year old man/woman who smokes, has obesity and diabetes will have the same likelihood of surviving free of disability as a 65 year old man/woman who is a non-smoker, of normal weight and does not have diabetes. This suggests that that the combined effects of smoking, obesity and diabetes biologically ages an individual by 20 years in terms of their likelihood of surviving free of disability over a 13-year period. From the standpoint of preventing disability in those who have combinations of potentially modifiable risk factors such as obesity, diabetes and smoking, we have now developed an algorithm that can estimate individual risks of disability, death and therefore survival free of disability. This algorithm therefore enables risk stratification of individuals and can be used as a tool by (1) the individual to drive motivation for modification of lifestyle, and (2) stakeholder groups such as workplaces, health insurers or government to link interventions by risk profile. In our studies to further our understanding of the obesity-disability relationships, we demonstrated that: 1. The simpler measures of BMI and waist circumference had the highest predictive ability for disability, thereby negating the use of more complex and expensive adiposity measures to predict the outcome of disability; and 2. With every additional year lived with obesity, there was a 3% increase in the risk of developing disability over and above the BMI attained. This finding supported the need for estimates of future health burden of obesity to consider the duration of obesity. Finally, we estimated that if prevalence of obesity and diabetes continues to increase from 1980 to previously predicted levels for 2025, there will be a 26% rise in disability prevalence after 2025 compared to disability attributable to 1980 obesity and diabetes prevalence. It is likely that this impact will be similar around the world in developed countries. It is imperative that when strategically planning to care for our ageing population, we take into consideration obesity and diabetes trends. Conclusion This body of work has added to the evidence base of the relationship between obesity, diabetes and disability. We have further demonstrated the additional impact of obesity duration on the obesity-disability relationship. Our novel risk algorithm identified obesity, diabetes and smoking as key modifiable risk factors in mid-life for disability in later life and demonstrated the combined effects of these risk factors on ageing. This disability risk algorithm has wide ranging applications including monitoring population health based on changes in risk factor profile of the population.


Principal supervisor

Anna Peeters

Year of Award


Department, School or Centre

Public Health and Preventive Medicine

Campus location



Doctor of Philosophy

Degree Type



Faculty of Medicine Nursing and Health Sciences

Usage metrics

    Faculty of Medicine, Nursing and Health Sciences Theses


    No categories selected


    Ref. manager