APPENDIX 3: DEVELOPMENT OF THE 1995 NNS–ADG DATABASE

Analysis of added sugars and ADG food consumption for the 1995 National Nutrition Survey (NNS) required the development of a new database of relevant ADG values for each food within AUSNUT 1999 (the classification used in the 1995 NNS). This 1995 NNS-ADG database was based on the 2011-13 AHS-ADG database developed by FSANZ for measuring ADG foods in the 2011-13 AHS (For more information, see Assessing the 2011-13 AHS against the Australian Dietary Guidelines - Classification System and Database Development Explanatory notes, available from:http://www.foodstandards.gov.au/science/monitoringnutrients/Pages/default.aspx).

The essence of the method was to assign the ADG values of foods to the AUSNUT 1999 foods from the 2011-13 database by finding/modifying suitable ‘donor’ records. The systematic process used to determine the most appropriate values from available information can be represented as a basic decision tree (see Figure 1) and is described more below.

Using the matching file

The first test was whether the 1999 food has a suitable match on the matching file produced and published by FSANZ. This was used a far as possible to concord AUSNUT 1999 codes to AUSNUT 2011-13 codes. In most cases, AUSNUT 1999 codes were matched to a single 2011-13 code, so these correspondences were used as is. In other cases, the matching file indicates that a single 1999 AUSNUT code had multiple 2011-13 matches (e.g. in 1999 there was a single ‘Soft drink, Energy’ code, however in 2011-13 this had increased to incorporate seven varieties). In these cases, an average was taken of ADG and added/free sugar information for all the possible matches and these values were assigned to the AUSNUT 1999 code.

Other assigning methods

Where the matching file did not specify a match for an AUSNUT 1999 food, the following approaches were taken in order:

• Manually assigned – foods that do not have a direct concordance to the 2011-13 AUSNUT file in the matching file were manually assigned a relevant 2011-13 AUSNUT food code ensuring the food was the closest in description and composition (including macronutrient profile).
• Recipes calculation– where a clear match could not be found and the 1999 AUSNUT food had a recipe, new ADG and free/added sugars values were derived by building up from the specified ingredients.
• Foods described as ‘not further specified (NFS)’ – values for the not further specified codes were calculated as the average ADG/added sugars values of the foods in that group. For example, data for ‘Potato, cooked, NFS’ was calculated from the consumption-weighted average of all the potato codes covering various preparation methods.

Added and free sugars

Added and free sugars were also previously not available for the 1995 NNS, so assigning values for these was done using the same method as for the ADG food amounts, with a further consistency step. Where the amount of total sugars was different between the 1995 food and its 2011-13 ‘donor’, the added and free sugar were adjusted to be the same ratio of free sugars to total sugars in the corresponding 2011-13 food. For example, in 2011-13 regular vanilla ice cream had 18.4g per 100g of total sugars and 13g per 100g of free sugars (ratio of 13.0/18.4). This ratio was then applied to the total sugars of regular vanilla ice cream in 1995 which was 19.8g, yielding 14g of free sugars.

Figure 1 — Decision tree for assigning ADG/added sugar values to 1999 AUSNT foods

For more information about determining the amount of added and free sugars, please refer to “Determining the amount of added sugars and free sugars in foods listed in the AUSNUT 2011-13 dataset” from Food Standards Australia and New Zealand website.

Applying the discretionary flag

Foods from the 1995 NNS were flagged as either non-discretionary or discretionary based on the same criteria used to flag foods from AUSNUT 2011-13.

Limitations

Care was taken to ensure every AUSNUT 1999 food was assigned to the most accurate values for the ADG food groups and free/added sugars possible. However, as the original AHS-ADG database was not designed specifically for the 1995 NNS data, the accuracy of the assigned ADG and added sugars information will be determined by a number of factors:

1. Changes in formulation of products. For multi-ingredient foods, specific individual ingredients may have changed. This may be the case for manufactured foods or home/commercially cooked foods where recipes have changed. On the other hand, even within a given period (whether 1995 or 2011-13) there may be variations in food composition because of factors such as production and processing practices, formulation changes, variations between brands and changes in the source of an ingredient.

2. Availability of a suitable match. While most foods in AUSNUT 1999 have direct counterparts in 2011-13, some values had to be imputed based on available recipes or as weighted averages of available similar foods where the 1995 food lacked detailed description (as described above). Such imputation is likely to introduce a degree of random error (compared to systematic the use of the comprehensive recipes used by FSANZ to create the 2011-13 AHS-ADG database). However, the impact on results from such random errors are considered to be limited and within the margins of sampling error when considering mean serves or grams of a food group at the population level.

3. The free/added sugars estimation has a further limitation with foods containing both intrinsic sugars (naturally present) and added sugars (for example, a particular yoghurt may have intrinsic sugars from the milk and fruit, but also be sweetened with sucrose). The limitation is due to the application of the 2011-13 ratio to the 1995 total sugars to determine the free and added sugars. While this ensures that the 1995 free and added sugars are adjusted in proportion with their amount total sugars, there may be instances where the difference in total sugars between a 1995 food and the 2011-13 food was due to an increase or decrease in either the intrinsic or the added sugar but not both (which is the assumption inherent in the use of the ratio).