The measures data in AUSNUT 2011-13 has undergone an extensive data validation process, with portions of the database validated through manual weighing of food and beverages by FSANZ.
Some of the routine validation activities undertaken by FSANZ include checking:
- the majority of densities are between 0.7 - 1.1 g/mL, except for high sugar syrups, very high salt foods, whipped or aerated foods, some powders, salad vegetables and some grated foods.
- the portion specific measures look consistent with those for related foods (e.g. the amount of different coffee types is similar when presented in the same type of cup)
- the densities look consistent within groups of like foods (e.g. all sugar sweetened cordial bases have a density of around 1.22)
- portion data increase in proportion to the increase in added foods, for example a burger with salad and cheese weighs more than a burger with salad.
FSANZ also undertook some more targeted validation activities including manually weighing the most frequently reported foods and measures or where the portion specific measures may have changed over time. Manual weighing focussed on the following food and container categories: takeaway cups, travel mugs, sandwich fillings, muffins, Subway, hamburgers, spreads, amount of milk added to hot drinks (e.g. splash and dash), commonly consumed fruit, vegetables and meat, mixed dishes, nuts, sushi, pizzas, spreads of margarine.
Uncertainty and limitations of food measures data
According to Rumpler et al (2008), errors in estimating the mass of a food consumed, represented one-third of the error in a 24 hour dietary recall and, together with the omission of foods, was the largest source of error in estimating food consumption and nutrient intake.
The density of a food not only depends on its chemical composition (i.e. how much salt, sugar, fat or ethanol it contains) but also, on the way it is cut, and packed into a container (for example, whether it is in a single large piece, chopped into large or small pieces, shredded, grated or pureed). Because we did not always know how the reported food was cut or packed there is considerable uncertainty associated with these values. Where this information was not available we have tried to provide a density that best represents how respondents were likely to have prepared and served foods.
Values for food specific portions may not fully account for all product sizes available. As with densities, values presented for portions represent the best estimate for the foods reported in the 2011-13 Australian Health Survey (AHS), particularly for most commonly consumed foods.
There are some foods for which it is difficult for respondents to clearly define size by providing dimensions. For example, it is not always easy to describe the length, width and depth of an irregularly shaped steak or chop, or the thickness of a pizza slice. In this situation, respondents were provided with the option to report the amount consumed using a general portion general descriptor such as a 'small steak'. These portion descriptors and gram amounts were generated by reviewing the size of products for sale at the time of the AHS, but have considerable uncertainty associated with them because of the subjectivity of the descriptors used.
Because the measures represent foods consumed during the AHS, the food specific portions may differ from portions reported in AUSNUT 1999 or AUSNUT 2007 because of changes in pack sizes, slice weights etc.