Food specific portions
Not further defined measures
FSANZ used a range of approaches for developing measures for foods and beverages consumed during the 2011‒12 National Nutrition and Physical Activity Survey (NNPAS) and 2012‒13 National Aboriginal and Torres Strait Islander Nutrition and Physical Activity Survey (NATSINPAS) components of the 2011‒13 Australian Health Survey (AHS).
Almost all foods and beverages in AUSNUT 2011–13 have been assigned at least one density, with units of grams per millilitre (g/mL). Where information on how the food was cut or prepared by the respondent was available, multiple densities have been assigned to the food. In this situation, FSANZ has provided an additional descriptor to enable users to select the appropriate density for their purposes. For example, for meats, the database contains a density for a solid piece of meat, to use for estimating the mass of a single piece of steak, and a density for chopped or diced meat, to use for estimating the mass when pieces are placed in a bowl or on a plate. In each case, we have tried to provide a density that best represents how respondents were likely to have prepared and served foods, but there is inevitably uncertainty associated with these values.
The majority of densities in AUSNUT 2011–13 were imputed from AUSNUT 2007 (FSANZ, 2008), developed to support the 2007 Australian National Children's Nutrition and Physical Activity survey (CSIRO, 2008) as this was the most up to date measures information FSANZ had available. The measures in this database underwent extensive validation before publication.
Other techniques such as analysis, borrowing values and, to a lesser extent, label data and estimation were also used.
In FSANZ's most recent analytical programs, including the Key Foods Programs (FSANZ, 2006; FSANZ 2008) and Australian Health Survey Programs (FSANZ, 2011) information on density (or specific gravity) for liquids such as milk, fruit juices and soft drinks was collected. Where available, these values have been incorporated into the AUSNUT 2011–13 food measures database.
Where Australian derived data was not available, data was borrowed from the United States Department of Agriculture (USDA) Post Interview Processing System (PIPS) measures database, developed to support the National Health and Nutrition Examination Survey (NHANES) and the USDA Standard Reference database (USDA, 2013). Values have only been borrowed where FSANZ was confident that the products are similar between countries.
Common cook books and online literature searches were also used to determine densities for foods commonly used as ingredients, such as a cup of flour.
Label information was used to determine the density of some breakfast cereals, particularly where a product label identifies a serving size in grams as a proportion of a cup (e.g. 'Average serve size = 45 g (¾ of a cup)'). Where this occurs FSANZ has calculated the density using the gram amount of the serve size and the volume of the portion of the cup identified (i.e. 45 g/ 187.5 mL = 0.24 g/mL).
Estimation has also been used to determine the density of some mixed dishes. This is because the density of these dishes is highly variable due to factors such as piece size, recipe proportions including the amount of sauce, and the degree of packing within containers. The densities of many of individual mixed dishes were determined analytically, with the values for similar style dishes (e.g. mixed dishes with sauce or with sauce and rice) being averaged.
Food specific portions
The majority of foods and beverages in AUSNUT 2011–13 have multiple measures for food specific portions. All data are for Australian foods and beverages and reflect sizes of products available for sale during the NNPAS, and may not reflect package sizes currently available.
A range of techniques were used to generate these measures. The most common techniques were analysis, estimation and label data, but some measures were based on borrowed data and industry data.
Many of the food specific portions were manually weighed by FSANZ, using household scales. This approach was mainly used to develop measures for commonly consumed foods and measures such as an apple or a potato. It was also used to determine a reasonable average value for foods and beverages with a wide range of serve sizes available such as coffees supplied in a range of takeaway cups; sandwich fillings; mixed dishes, takeaway and fast food products reported as bowls, mounds and takeaway containers; the quantity of milk added to tea and coffee and of spreads such as vegemite added to bread; and portions of commonly consumed meats, seafood, fruit and vegetables.
For more information on a selection of the programs undertaken by FSANZ to develop food specific portions during the AHS, select the relevant program listed below.
Estimation has been used to describe a wide range of approaches for generating measures for food specific portions. For example, it has been used to calculate the gram amount of:
- Single cooked foods using the raw weight of the food, usually derived by analysis, and a known weight change factor. This approach was most commonly used to determine measure data for toast, and cooked eggs, meat, seafood and vegetables.
- Multiple ingredient foods consumed using the mass of the ingoing ingredients, any relevant weight change factors and the average number of suggested serves identified by the recipe (if relevant). This approach was most commonly used to determine measure data for cakes, slices and other desserts, takeaway foods such as hamburgers, hot dogs, sandwiches etc
- A food or beverage consumed using a known volume and the relevant density. This approach was most commonly used to determine the measure data for sauces, milkshakes, thickshakes, bites, mouthfuls and margarine spreads.
- A food using the edible and inedible portion. This approach was mainly used to determine the weight of cuts of meat with different levels of trimming and for fruits and vegetables to account for the removal or addition of peel.
- Wild caught and harvested foods consumed during the NATSINPAS such as echidna, turtle and yam where limited data was available on typical portion sizes.
- 'Not further defined' measures. This occurred where a respondent was unable to identify the exact quantity of food or beverage they consumed.
Estimation was also used to determine the best estimate where a range of analytical and/or label values were available or where limited data was available and where limited measures were available such as for wild caught foods and the choice of measure size was unlikely to have any significant effect on the AHS outcomes.
This technique was mainly used to develop measures for foods and beverages where a brand was identified by respondents, although it has also been used to develop measures that reflect commonly consumed commercial products where a brand was not identified.
Label information was most commonly used to develop measures for sweet and savoury biscuits, breakfast cereals, fast food products, yoghurts, confectionery, frozen meals, muesli bars, chips and other snack foods, and other commonly consumed commercial individual serve items.
Some measures have been imputed from AUSNUT 1999 (ANZFA, 1999) and AUSNUT 2007 (FSANZ, 2008) or borrowed from the USDA PIPS measures database or Standard Reference Database. These techniques have only been used where FSANZ has confidence in the validity of the assumptions made.
A small number of measures were also based on values provided by the food industry. For example, strawberry measures were based on the average weight of small, medium and large strawberries supplied by the Victorian Strawberry Growers (2006); egg measures were based on the average weight and edible portions of different size eggs provided by the Australian Egg Corporation (2007); and the density of diced and sliced beef was based on unpublished analytical work undertaken by the Meat and Livestock Australia. Industry data was only used where FSANZ had confidence in the data provided and its relevance to the way measures have been reported in the AHS.
'Not further defined' measures
Not further defined measures were developed for survey foods and beverages where a respondent was unable to identify the exact quantity of food or beverage they consumed. For example, a respondent might have reported they drank water, but did not indicate how much water they drank.
Not further defined measures were derived using two approaches:
- Using the average serving size as reported in the NNPAS or NATSINPAS. This approach was most commonly used for beverages, mixed dishes and commonly consumed vegetables. In some instances the eating occasion also needed to be taken into account. For example, for beverages, different measures were created based on whether the eating occasion was flagged as a beverage (i.e. a single drink) or extended consumption (i.e. consumed over the course of the day). For some foods, particularly main meal dishes, consideration was given to the age of the respondent consuming the food, with smaller values generally assigned to meals eaten by young children.
- Assigning an undefined measure the most frequently consumed or most likely portion size. For example, an unknown measure for the food Beef, steak, semi-trimmed, baked, roasted, fried, grilled or BBQ'd, no added fat could be assigned the measure for steak, medium as it most commonly eaten as a whole steak and a medium size was assumed to be the average size eaten.
In some instances, a food has been assigned two ‘not further defined measures’ depending on whether the measure was used in the NNPAS or the NATSINPAS. Where two measures exist, the food has been assigned a flag to help users identify which part of the AHS the measure was used in.
Due to the weighting techniques used to develop undefined measures, the resulting data do not reflect any particular food consumed. In addition the weighting assumptions reflect patterns observed in this AHS and may not be appropriate for use in other circumstances. In particular, the food specific portions developed for this survey will not be able to be directly compared to those developed for the 1995 NNS and as such cannot be used to analyse changes in portion sizes over time.