Background: Portion size estimation is one of the largest sources of error in dietary assessment and relies on individuals’ perception, memory and conceptual skills(1). Varying methods adopted by researchers and practitioners have led to conflicting conclusions on accuracy of current dietary assessment methods including food photography and household measure estimations. There is also an absence of research focusing on young adults in this area. This study aimed to assess the accuracy of portion size estimations using household measures and food photographs compared to actual weights, among young adults.
Methods: In a cross-sectional study design, 35 (18 female, 17 male; aged 18–26 years) participants were recruited by posters distributed around a North England university campus. Participants volunteered by arriving at a designated location indicated on the posters. A short demographic questionnaire was completed before participants self- served a meal consisting of three items with either a definite shape (jacket potato) or amorphous foods of no defined shape (baked beans and grated cheese). Serving dishes were weighed before and after serving, providing the actual weight of foods served. Fifteen minutes after eating, a questionnaire was used to collect dietary information. Participants described portions using household measures e.g. number of spoonfuls or ‘small’, ‘medium’ or ‘large’. These were quantified using a portion size reference guide(2). A food photograph atlas(3) was used for selection of a photograph representing each item and the associated weight recorded. In total each participant provided three weights per component (i.e. actual weight, household measure estimate, food photograph estimate) providing 315 weights overall.
Results: For the overall meal, household measure estimations were the most accurate with an average 4% underestimation compared to actual weight. Food photographs overestimated portion weight by 14% on average. For individual meal components food photographs were more accurate in estimating the size of jacket potatoes with no significant difference between estimates and actual weights (P = 0.34). Cheese was significantly overestimated by food photographs (P = 0.029) and underestimated by household description (P = 0.005) compared to actual weights. Baked beans were overestimated by both methods. Females overestimated more frequently, however differences in estimation between males and females were not statistically significant.
Discussion: Estimated portion weights of all three food components differed significantly to actual weight for at least one of the methods. It would therefore be inappropriate to suggest that either method can assess the dietary intake of young adults without error. Household measure descriptions were closer to actual weight than estimations based on food photographs. Amorphous foods may be particularly prone to overestimation regardless of method. Consideration should be given to the likely overestimation in portion size associated with food photographs and underestimation with household measures. Both under- and over-estimation can be associated with household measure descriptions according to the type of food being assessed. Food photographs and household methods are useful portion size assessment tools for use among young adults only when inherent errors in both methods are acknowledged.
CORE (COnnecting REpositories)
CORE (COnnecting REpositories)