Prevalence and characteristics of energy intake under-reporting among Australian adults in 1995 and 2011 to 2012.

Aim Misreporting of energy intake is a common source of measurement error found in dietary surveys, resulting in biased estimates and a reduction in statistical power. The present study aims to refine the conventional cut-off methods and to examine the extent to which Australian adults misreport their energy intake, and the characteristics of under-reporters between two time points. Methods A revised Goldberg cut-off approach was used to identify those who reported implausible intake amounts in a secondary analysis of two large cross-sectional surveys. Identified low energy reporters were then used as the outcome variable in Poisson regressions to examine association with sex, age, body mass index (BMI), weight perceptions, education, relative household income, geographic remoteness and relative socioeconomic disadvantage. Results The prevalence of under-reporting increased from 32% in 1995 to 41% in 2012, most of which can be attributed to an increase in men. Under-reporting has a positive association with BMI and relative socioeconomic disadvantage, but an inverse association with age, education, relative household income and residence in inner regional areas. Conclusions Under-reporting of energy intake is high in Australian adults, and appears have worsened over time in men, which could be partly explained by the upward trend in obesity. The use of conventional Goldberg method to identify under-reporters can greatly underestimate the prevalence of under-reporting, future studies should consider selecting a lower critical value to improve accuracy.