Accounting for misreporting when comparing energy intake across time in Canada.

BACKGROUND

Estimates of energy intake are lower in 2015 compared with 2004. The difference observed is too large to be explained by a change in energy requirements or physical activity at the population level. Self-reported dietary intake is subject to misreporting and may explain part of this difference. The objectives of this study are to assess how misreporting has changed from 2004 to 2015 and to demonstrate how these changes may affect the interpretation of the national intake data of Canadians.

DATA AND METHODS

Data from the 2004 Canadian Community Health Survey - Nutrition (CCHS - Nutrition) and the 2015 CCHS - Nutrition were used to estimate energy intake and requirements for all participants aged 2 or older. The ratio of energy intake to total energy expenditure requirements (EI:TEE) was used to categorize respondents as under-reporters (EI:TEE ⟨ 0.70), over-reporters (EI:TEE ⟩ 1.42) or plausible reporters (EI:TEE = 0.70 to 1.42). Descriptive analyses by category of respondent were conducted for respondents aged 2 or older who participated in the measured height and weight component. The main caloric sources that contributed to the difference in estimated energy requirements were used to show the impact of misreporting on the analysis.

RESULTS

The prevalence of under-reporters was 7.5% higher in 2015 compared with 2004, while the prevalence of over-reporters was 7.4% lower. There was no change in the prevalence of plausible reporters. Estimated energy intake from participants categorized as plausible reporters showed a difference of 84 kcal from 2004 to 2015, compared with a difference of 250 kcal for the entire sample. Estimated energy intake was lower in 2015 compared with 2004 across all categories of respondents for many foods, including sugar-sweetened beverages and milk, and was higher for only pastries and nuts.

DISCUSSION

Misreporting changes will affect analysis and should, at a minimum, be acknowledged when comparing 2015 with 2004. Using a comparable category of plausible reporters or adjusting for reporting status are options that will allow a better comparison of these two datasets.