Degree Name

Doctor of Philosophy


School of Health Sciences


Introduction: Ensuring good health and wellbeing into old age is essential as the population is ageing at a rapid rate and the number of older adults aged 65 and over will make up almost a quarter of the Australian population by 2036. Older adults are at greater risk of developing chronic diseases and suffering functional limitations as they age. Therefore appropriate services are required to help prevent and/or manage these issues so people can remain living longer and disease free in the community. A number of these issues affecting older adults are known to be diet related. However, a healthy diet is difficult to achieve if people do not have sufficient access to, and availability of, nutritious foods, i.e. they are food secure. National measurement of food security in Australia has been limited to a single item tool and consequent reports suggest only 5% of the population run out of money and do not have enough to buy more food. The single item does not explore the anxiety about acquiring food, or the quality or quantity of food, but is limited to a brief assessment of the economic ability to acquire food. If older adults are unable to access appropriate foods for good health then their overall diet quality could be affected. Evidence from other developed countries has found significant relationships between poor diet quality and mortality as well as increased risk of poor health. Diet quality indices to measure overall diet in terms of adhering to recommended dietary guidelines have been developed over the past few decades; yet research in Australia has been limited and findings reported primarily from cross sectional studies.

This research has two main aims: Firstly to determine the prevalence of food insecurity and changes over ten years in a cohort of older Australians using a comprehensive food security survey. Secondly to investigate whether older Australians adhere to recommended dietary guidelines using a newly developed diet quality index.

Methods: This thesis comprises five studies based on secondary data analysis from The Blue Mountains Eye Study (BMES), a population based cohort of older adults aged 49 years and older at baseline. Between 1992 and 1994 3654 (87.9% response rate) participants were enrolled at baseline (BMES1) and were invited to attend follow up visits every five years (BMES2, BMES3 and BMES4). At baseline and each followup assessment, participants completed in depth questionnaires that included demographics, self-reported medical history and a 145-item food frequency questionnaire. A comprehensive 12-item food security survey was included in the BMES2 and BMES4 follow ups. Prevalence of, risk factors for and change over a 10 year period in food insecurity status were investigated using logistic regression models.

Diet quality was indicated by adherence to the Dietary Guidelines for Australian Adults. A new tool, the Total Diet Score (TDS) was modified and adapted from a previous Australian diet quality index project and followed the US 2005 Dietary Guidelines Adherence Index. The tool was divided into ten components each of which had equal weighting and a maximum score of 20 was achieved if an individual adhered closely to the dietary guidelines. The relationship between TDS and allcause mortality was examined using Cox proportional hazard risk ratios.

To measure the accuracy of the TDS, a validation study was undertaken by comparing TDS scores calculated from the FFQ to TDS scores calculated from weighed food records completed by 75 participants from BMES1. In addition, the relationship between TDS scores and nutritional biomarkers were assessed by analysis of covariance in the BMES2 Cross Section cohort.

Findings: The prevalence of food insecurity was 13% in the BMES2 Cross Section, considerably higher than previously reported using the single item economic tool. Risk factors for reporting food insecurity included being female, less than 70 years old, living in rented housing and living on a pension. Food security remained an issue when participants enrolled in BMES2 and BMES4 were examined at ten years (BMES4). In the survivor cohort, 4.8% reported being food insecure if previously food secure and 3% reported being food insecure at both time points. The risk factors for food insecurity over time included being divorced/separated, obese and living on a pension only. Loss to follow up was an issue in this cohort and predictors of death prior to BMES4 included being male, up to high school education, widowed, renting a home, diabetes, acute myocardial infarction, walking disability and poor self-rated health.

To address the second aim, the Total Diet Score was used to determine the relationship between diet quality and all-cause mortality. Participants in the highest quintile of diet quality had a 21% reduced risk of all-cause mortality compared to those in the lowest quintile of diet quality (HR 0.79, 95% CIs: 0.63-0.98, ptrend = 0.04). As was expected the consumption of optimal choice foods increased with increasing TDS quintile; for example dairy intake increased from quintile 1 to quintile 5 by 31%, but low fat or reduced fat milk increased by 224% across the quintiles. The TDS scores from the FFQ were not significantly different from WFR TDS scores. The regression analysis also found no significant linear trend between the differences and the means of the TDS scores from the FFQ and WFRs using the Bland Altman method.

No relationship was found between food security and diet quality nor was diet quality a mediator between food security and health related quality of life. Quality of life, measured through the SF-36 Health Survey, was found to be significantly lower across all domains of physical and mental health in food insecure individuals compared to food secure individuals after adjusting for a range of covariates. Lower adjusted mean SF-36 scores were also found in the lower quintiles of diet quality than higher quintiles of diet quality; however the differences were smaller and not significant.

Conclusions: This study found rates of food insecurity in an older Australian population were higher than previously reported. This was also the first study to investigate changes over time in an Australian population. The higher prevalence rate of food insecurity found in this study was determined through the use of a comprehensive measurement tool addressing anxiety, quality and quantity of the diet. Such a measure of food insecurity is in line with current food security definitions unlike the single item tool that only asks about the economic ability to access food. Of particular importance was the finding that a small number remained food insecure after ten years and almost 5% of older adults became food insecure.

Diet quality in an older population was low but those with a higher diet quality had a reduced risk of mortality, suggesting efforts are required to improve the diets of older adults to improve their health outcomes. The TDS was a valid tool for measuring overall diet quality in this population. It could be adapted for use in other age groups by adapting serve sizes in the scoring protocol. Further research is required in this area to adapt the TDS to the current dietary guidelines with appropriate validation techniques to ensure the TDS measures overall diet quality as accurately as possible.

However, improving diet quality and health outcomes in older adults may only be achieved if they are food secure and have the ability, economically and physically, to access food as was shown by the significantly lower physical and mental quality of life scores in food insecure individuals. Therefore, food insecurity monitoring and measurement utilising an appropriate tool is required in Australia as a public health priority.



Unless otherwise indicated, the views expressed in this thesis are those of the author and do not necessarily represent the views of the University of Wollongong.