Validating a nutrient profile model

Abstract

Objective: To assess the validity of nutrient profiling Model WXYfm - developed for the purpose of regulating the promotion of 'less healthy' foods to children. The model ranks foods according to their healthiness and categorises foods into 'healthier' and 'less healthy' foods.

Design: Convergent and discriminant validity was tested by comparing the way Model WXYfm categorises foods with the way the UK's national food guide - the Balance of Good Health (BGH) - categorises foods. Construct validity was assessed by testing a hypothesis relating the constructs of 'healthiness' of foods (as measured by Model WXYfm) and the 'healthiness' of diets (measured using the Diet Quality Index) and assessing whether this hypothesis was confirmed or refuted by using data on the dietary patterns of subjects (n = 1117) of the National Diet and Nutrition Survey of adults carried out in Great Britain in 2000-01.

Results: Model WXYfm showed good convergent and discriminant validity: the level of agreement between the way the model categorises foods and the way the BGH categorises foods was good (kappa = 0.69). Model WXYfm also showed good construct validity: the energy intake from 'less healthy' foods amongst subjects with the least healthy diets was nearly twice the energy intake from 'less healthy' foods amongst the subjects with the healthiest diets.

Conclusions: Model WXYfm demonstrated good validity in categorising foods in a way that is related to the healthiness of diets both recommended and achieved. The methods for assessing the validity of a nutrient profile model used in this paper have not, to our knowledge, been used before.