Accuracy of customers’ caloric estimates of ultra processed foods

Introduction: Ultra-processed foods (UPF) have been defined as containing multiple ingredients and high levels of consumption are associated with higher intakes of energy and an increased likelihood of obesity. Perceptions of these foods, including estimates of their caloric content, is key to explore given their possible influence on consumers’ food choices.

Aim: The aim of this research was to investigate the accuracy of consumers’ caloric estimates of UPF and compare these with those for non-UPF products while exploring potential consumer and product-level sources of bias.

Method: An online image-based questionnaire was developed and used to obtain respondents’ calorie estimates for four pairs of UPF/Non-UPF, from the grains, meat, starchy vegetables and diary food groups, which were presented as individual servings without label information. The questionnaire was completed by a convenience sample of consumers who were aged 18yrs+ and resident in the UK or The Netherlands. The accuracy and variance of respondents’ average calorie estimates was calculated in relation to each food’s actual energy content (kcal), and reported together with ± standard deviations (SD). Analysis was undertaken using t-tests to detect differences in the accuracy of caloric estimates within pairs and in relation to respondents’ characteristics (i.e. when sub-grouped according to frequency of nutrition label usage and levels of product-specific familiarity).

Results: Data from 112 respondents were included in the analysis (mean age: 35,8 years ± 16.3, female 66.1%, UK resident 29.5%). The accuracy of respondents’ average caloric estimates varied across UPF (milkshake +44.3kcal ± 123.3kcal, breakfast cereal +39.5kcal ± 110.4kcal, chicken nuggets +51.6kcal ± 140.4kcal, French fries +64.0kcal ± 165.9kcal) and Non-UFP (milk -1.8kcal ± 78.4kcal, porridge -42.4kcal ± 113.3kcal, chicken breast fillet +51.6kcal ± 51.6kcal, baked potatoes +76.4kcal ± 140.5kcal). Respondents were more likely to overestimate the calorie content of the milkshake, breakfast cereal and chicken nuggets than their corresponding Non-UPF products; milk, porridge and chicken breast (all p=<0.001).

Conclusion: Data from this convenience sample suggests consumer estimates of both UPF and Non-UPF products, vary in their accuracy, but overestimation of the caloric content of some UPF products is also likely. Further work exploring the consumer and product-level factors which influence calorie content estimation of UPF is warranted, including in relation to perceived product healthfulness.