Degree Name

Doctor of Philosophy


School of Medicine


Background This thesis focuses on dietetic practice supporting weight loss. Whereas dietary intervention research forms the basis for evidence based practice, the focus on nutrients, ingredients or single foods may limit the conclusions that can be drawn for the practice setting. In practice, dietitians assess and provide the dietary prescription only in terms of foods, not nutrients. Observational research has been conducted in terms of foods and dietary patterns however this does not necessarily translate to advice for individuals and the findings of this research needs to be confirmed with randomised controlled trials. Although reducing energy intake relative to expenditure is pivotal in weight loss, questions remain about which foods and dietary patterns optimise weight control. Weight loss is complex, and diet quality may be overlooked in attempts to lower kilojoule intake. A healthy diet can be constructed in numerous ways, but certain features are thought to be important for health and linked to better health outcomes. The foods consumed within the whole diet change over time, but analysis of these changes, particularly in the context of an intervention, may help clarify the most effective dietary changes and provide more specific food advice, augmenting nutrient‐level findings. This research explores a whole‐of‐diet approach where categories of foods were used to monitor dietary change and weight loss in a clinical intervention context forming links between dietary patterns and outcomes.

Research Hypothesis and Aims The central hypothesis examined in this thesis was that an analysis of dietary patterns reported by overweight participants in a weight loss trial will reveal important practice‐relevant information for developing dietary advice.

The aims of the thesis were to: 1. Develop food categories extending from the traditional five food groups to explore dietary patterns within an intervention context. 2. Identify patterns of food choice in the context of a clinical weight loss trial and associations with weight loss and health outcomes. 3. Develop and validate a diet quality tool for weight management, to investigate changing diet quality within a weight loss intervention context.

Methods This thesis involved a number of investigations using secondary analysis of combined data from two dietary intervention trials [1, 2] conducted through the Smart Foods Centre at the University of Wollongong. Both intervention studies measured a primary outcome of weight loss. For this thesis, to further examine the nutrition‐health interface, methods for considering whole foods within the whole diet needed to be developed. This involved developing a system for categorising foods, methods for distilling dietary patterns and approaches to examining links between dietary patterns and outcomes.

Developing food categories for the examination of dietary patterns was central to the thesis framework. Defensible food categories were derived from a review of the literature on associations between food intake and health outcomes, an examination of existing food categorisation systems and a consideration of the culinary use of food. Cluster analysis was utilised to examine dietary data from these new categories at baseline, and the association between weight loss and changing patterns after three months. Dietary modelling was applied to two idealised energy deficit diet models (6500 and 7400kJ) in the development and validation of a diet quality tool, referred to as the Food Choices Score (FCS). The diet models assured adequate nutrient intake based on core food groups. Using this score, dietary patterns were examined between baseline and three months. Consumption of certain food categories (particularly noncore foods and drinks) were examined for up to 12‐months in relation to weight change. Comparisons to computerised nutrient analysis, biochemical and anthropometric measures using data at baseline, three months and 12‐months were conducted.

Results Seventeen food categories were derived with some reference to the traditional five food groups to clearly depict dietary patterns. The 17 categories assisted in provision of more descriptive information in assessing the associations between dietary patterns and positive and negative health outcomes than the traditional five food groups. While energy intake was significantly reduced within three months, the weight of food (excluding fluids) was not reduced. This result was primarily due to reductions in consumption of higher energy food choices, non‐core foods and drinks (NCFD), fatty meats and non‐wholegrain cereal foods and a corresponding increase in vegetable consumption. Cluster analysis using the same data showed subjects consuming >6 serves of NCFD at baseline lost significantly more weight than those with baseline diets already closer to dietary targets for weight loss. A higher FCS was representative of higher diet quality and a greater improvement in the FCS over three months predicted greater weight loss validating it as a useful tool. The changes in diet patterns in relation to weight loss were further confirmed through logistic regression reinforcing the link with decreasing consumption of NCFD and non‐wholegrain cereal foods. Dietary patterns of participants with weight loss >10% over 12‐months reported consuming significantly less NCFD and significantly more fruit than those losing <5%.

Major conclusions and relevance to dietetic practice The analysis based on reported food consumption provided insights in terms of specific food‐based dietary advice. The methods employed for examining dietary patterns maintained the detail of the foods described in the original dietary data. The outcomes of this thesis suggest that food‐level analyses should accompany analyses of energy and nutrient intakes in the routine examination of dietary changes in dietary intervention trials. The addition of dietary patterns would help facilitate translation of research findings to the clinical setting. In the analysis presented, weight loss was greater in those with higher baseline intakes of NCFD. Thus close examination of the detail of food choices at baseline may help maximise the energy deficit that could be created for individuals desiring weight reduction. In contrast, reporting a dietary pattern closer to guidelines at baseline may mean fewer dietary changes, however reducing energy intake may still be required. Intakes of some foods and drinks may be more modifiable than others during weight loss. In the analyses reported in this thesis, large reductions in energy intake were possible from foods categorised as NCFD. Monitoring the intake of all foods and drinks over a longer time‐frame was shown to be important in monitoring body weight change, especially since the energy density of some foods may influence the amount of total food ingested. Food weight was consistent across the first 3‐months, although mean energy decreased >2000kJ, presenting an opportunity for clinicians who could replicate the substitution of light weight yet energy dense foods with relatively heavy foods like vegetables. Diet quality tools, such as the Food Choices Score (FCS) developed specifically for the weight loss setting, may assist in clinical research by providing an opportunity to benchmark the baseline diet against ideals over time. This helps in focusing both the clinician and the client on the issue of diet quality in weight loss and thereby more effective food substitution.

Conclusions The analysis of change in food choices and dietary patterns at the dietary intervention level was novel and informative for practice. Using the 17 defined food categories proved useful for food pattern analysis and gave a meaningful representation of the reported diet history. Baseline dietary patterns particularly are a significant consideration in correcting dietary exposure for weight loss and a validated Food Choices Score was sensitive to dietary change in a weight loss context. A food‐based approach is a valuable adjunct to other analyses in weight loss trials and provides more specific food advice for the practice setting. The 17 food categories developed in this thesis provided a framework for further research at the dietary intervention level.

FoR codes (2008)




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.