A meta-analysis of children’s self-reports of dietary intake

Abstract

Objective: Although research studies increasingly use children as primary reporters in dietary assessments, it is unclear how well children’s self-reported intake correlates with independently validated reports of their intake; this meta-analysis assesses that correlation.

Design: Moderators of the correlation between self-reported and independently validated intake were predicted a priori: type of dietary intake assessment (24 h recall, food diary and food frequency questionnaires), validation measures, parental assistance and age. Online databases were searched for articles published from 1990 to 2014 that compared children’s self-reports of dietary intake to validated observations of food intake in children age 4–16.

Main outcome measures: Summary effect size Pearson r between children’s self-reported dietary intake and independently validated dietary intake were calculated.

Results: In k = 32 samples from 23 studies, a statistically significant correlation (r = .48, Z = 7.26, p < .001) was found between children’s self-reported dietary intake and independently validated reports of dietary intake. Validation method (Q = 17.49, df = 2, p < .001) and parental assistance (Z = 2.03, p = .042) were significant moderators of this correlation. Self-report methodology (Q = 3.95, df = 2, p = .139) and age (Q = .02, p = .879) were not significant moderators of the distribution of effect sizes.

Conclusion: Together, these results provide baseline information about children’s recall in dietary intake assessments conducted with children as primary reporters.

Keywords:

• meta-analysis
• dietary intake
• children
• self-reports
• validation studies

Notes

1. Although differences were not statistically significant, the correlation between self-reported intake and independent validation was lowest with the FFQ (r = .27), greater in 24 h recall (r = .48), and highest in food diaries (r = .57). See Table 2 for more detail.

2. Although differences between age groups were not statistically significant, the correlation between self-reported intake and independent validation was lowest with 8–12 years of children (r = .46), greater in 12–16 years of children (r = .48) and highest in 4–8 years of children (r = .54). See Table 2 for more detail.