Does Providing Assistance to Children and Adolescents Increase Repeatability and Plausibility of Self-Reporting Using a Web-Based Dietary Recall Instrument?

BACKGROUND: It is important to find ways to minimize errors when children self-report food consumption.

OBJECTIVE: The objective of this study was to investigate whether assistance given to children completing a self-administered 24-hour dietary recall instrument called SACANA (Self-Administered Child, Adolescent and Adult Nutrition Assessment) increased the repeatability and plausibility of energy intake (EI) estimates.

PARTICIPANTS/SETTING: The study was conducted between October 2013 and March 2016 in a convenience sample of 395 children, aged 8 to 17 years, from eight European countries participating in the I.Family study.

DESIGN: SACANA was used to recall the previous day's food intake, twice in a day, once with and once without assistance.

MAIN OUTCOME MEASURES: The difference in EI between the first and second recalls was the main repeatability measure; the ratio of EI to basal metabolic rate was the plausibility measure.

STATISTICAL METHODS: Generalized linear mixed models, adjusted for sex, age, and body mass index z-score, were used to assess whether assistance during the first vs second recall influenced repeatability and plausibility.

RESULTS: The difference in estimated EI (EI from second recall minus EI from first recall) was significantly lower (P<0.001) in those assisted at first (median=-76 kcal) than those assisted at second recall (median=282 kcal). Modeling showed that EI at assisted first recall was 19% higher (95% CI 1.13 to 1.24) than in assisted second recall. Overall, 60% of recalls had a plausible EI. Modeling to estimate the simultaneous effects of second vs first recall and assistance vs no assistance on plausibility showed that those assisted at first recall had significantly higher odds of a plausible recall than those unassisted (odds ratio 3.64, 95% CI 2.20 to 6.01), with no significant difference in plausibility of second recall compared to the first (odds ratio 1.48, 95% CI 0.92 to 2.35).

CONCLUSIONS: When children are assisted at first recall, the plausibility and repeatability of the later unassisted recall improve. This improvement was evident for all ages. A future, adequately powered study is required to investigate the age range for which assistance is advisable.