Assigning foods and portions reported to AUSNUT 2011-13 data

Each food consumed in the 2011-13 Australian Health Survey (AHS), is coded to a nutrient profile, which has a 5-digit food classifcation code associated with it and one or more measures to enable the gram amount of food consumed to be calculated.

General process for assigning food and portion information

FSANZ worked with the Australian Bureau of Statistics (ABS) to ensure each food and portion reported was coded to an appropriate food descriptor with an associated nutrient profile and measure descriptor in AUSNUT 2011-13. Foods and portions were coded using the key steps outlined below.

For more information on how food and measures were coded during the AHS refer to the ABS AHS User Guide.

Assigning foods consumed to AUSNUT 2011-13 nutrient data

Due to the large number of respondents and wide variety of foods available in Australia it is not possible to produce a nutrient profile for every unique food reported in the AHS.

Where possible, the ABS and FSANZ coded foods reported to a food with a very similar nutrient profile in the core dataset. Where this was not possible, new nutrient profiles were created to reflect the food consumed.

A new nutrient profile was not created to reflect every attribute reported. Before a new nutrient profile was created, consideration was given to:

• how different the nutrient profile of each food reported was compared with nutrient profiles in the core dataset and other similar foods reported
• how frequently the food was consumed.

A range of approaches were used to minimise the number of foods in AUSNUT 2011-13, while maintaining nutritionally relevant information. For example:

• Foods reported that had a very similar nutrient profile were grouped, with a single nutrient profile developed to reflect the average composition of the range of products identified. This appropach was commonly used for coding foods reported by brand. For example, the food Bread, from white flour, commercial has been used to represent the nutrient profile of many brands of commercial white bread reported.
• Similar cooking methods reported were grouped, unless analytical data were available for the specific cooking methods identified. For example, all meat, poultry, fish or vegetables described by the respondent as being 'baked', 'roasted', 'grilled' or 'BBQ'd' have been coded to the same nutrient profile.
• Combining the different types of fat reported. For example, although information was collected on the type of fat used to prepare a cake or to fry a piece of meat, a separate nutrient profile was not always created. If a food was frequently reported (such as chicken breast) a nutrient profile may have been created to capture the specific type of fat used (e.g. Chicken, breast, flesh, baked, roasted, fried, stir-fried, grilled or BBQ'd, olive oil). But for less commonly reported foods, a single nutrient profile was generated using an undefined fat that was weighted according to the fat types reported for the relevant food category.
• Grain types reported in mixed dishes were grouped. Although information was collected on all the grains eaten as part of a mixed dish such as a casserole, a single nutrient profile of Grains, boiled, for homemade mixed dishes was used in all these recipes, and its nutrient profile was generated using a recipe based on the frequency of consumption of all grains added to mixed dishes, as reported in the AHS.
• Nutrient profiles were not created for infrequently reported foods, unless they were a likely to be a major contributor to the intake of a particular nutrient. Instead, the food was coded to the food with the closest nutrient profile or a 'not further defined' food representing the weighted average of all foods in that category.

This process resulted in over 167,000 unique foods reported during the NNPAS being reduced to just over 5,700 foods. In a large scale population survey such as the AHS, these assumptions are unlikely to have a significant impact on population nutrient intake estimates.

New nutrient profiles were generally created to reflect differences in:

• home-prepared and commercial products
• species and varieties - for example fish species and fruit and vegetable varieties
• meat cuts and trimming practices
• raw and cooked foods
• dry cooked (e.g. grilled) and wet cooked (e.g. boiled) foods
• foods cooked with fat and without fat
• fortified and unfortified products
• foods containing added sugar and intense sweeteners
• presence of caffeine
• regular salt and reduced salt products
• regular fat and reduced fat products
• preparation practices and consumption patterns observed between the NNPAS and the NATSINPAS. 

These factors are likely to have important impacts on intake estimates for the nutrients being reported in the AHS.

The approach used to coding foods and portions evolved as the AHS progressed and some food categories, where coding occurred later in the AHS cycle (e.g. pasta dishes), may contain less detail than those developed earlier (e.g. cakes).

Some foods were also created specifically for use in the NNPAS or the NATSINPAS. A 'survey flag' has been assigned to foods to help users identify which component of the survey the food was used in. If the field is left blank then the food is relevant to both components of the AHS.

Assigning food portions consumed to AUSNUT 2011-13 measures data

After a reported food was coded to a FSANZ food descriptor, it was then assigned a FSANZ measure descriptor that reflected the portion reported.

Where possible, the ABS and FSANZ coded portions reported to an existing measure descriptor. Where this was not possible, new measures were created to reflect the portion consumed. A new measure was not created to reflect every unique portion reported.

Given the variation of portion sizes available for some foods  it is not possible to account for all product sizes available. FSANZ often needed to determine a reasonable average value for foods and beverages with a wide range of serve sizes available. For more information refer to the development of the food measures database.