Defining 'low in fat' and 'high in fat' when applied to a food
Arambepola C., Scarborough P., Boxer A., Rayner M.
Objectives: To describe four different methods of identifying indicator foods that are high, medium or low in fat with reference to dietary patterns and to use these indicator foods to test three sets of definitions of 'high', 'medium' and 'low' in fat from 'banding schemes' developed by the Coronary Prevention Group (CPG), the Food Standards Agency (FSA) and Sainsbury's. Methods: Indicator foods were developed using food intake data from the UK National Diet and Nutrition Survey and two parameters: (i) probability of the food being consumed by an individual with a high-fat diet (Method 1); and (ii) the contribution of the food to the fat intake of the average diet of consumers (Methods 3 and 4). Method 2 used both parameters. The three banding schemes were tested by assessing their levels of agreement with methods in categorising indicators. Results: Sensitivity in identifying high, medium and low fat indicators was highest with the CPG banding scheme (high and medium fat indicators) and Sainsbury's scheme (low fat indicators) (Methods 2, 3 and 4). The levels of agreement (kappa coefficient) were 0.68 for the CPG scheme; 0.51 for the Sainsbury's scheme; and 0.41 for the FSA scheme (Method 3). Conclusions: It is possible to use indicator foods related to dietary patterns of a specific population to generate more rational definitions of 'high', 'medium' and 'low' in fat. This could be the starting point for the development of indicator foods for testing more complex nutrient profile models (i.e. those that consider more than one nutrient). © The Authors 2008.