To assess the validity of the Food Intake Recording Software System (FIRSSt) against observation of school lunch and a 24-hour dietary recall (24hDR); and to test the effects of sequencing, observation and a hair sample as a bogus pipeline on accuracy of dietary report.
Six-group design systematically varying sequence of self-report (FIRSSt vs 24hDR), observation of school lunch and hair sample as a bogus pipeline manipulation, with random assignment of participants.
138 fourth-grade students in 2 elementary schools.
MAIN OUTCOME MEASURES:
Accuracy of reported food consumption was measured in terms of matches, intrusions, and omissions among the FIRSSt, 24hDR, and as observed at school lunch. Students also completed self-report of performance with FIRSSt.
STATISTICAL ANALYSIS PERFORMED:
t tests, Pearson correlations, analysis of variance, factor analysis.
When compared with school lunch observation for one meal, FIRSSt attained 46% match, 24% intrusion and 30% omission rates, while a dietitian-conducted 24hDR obtained 59% match, 17% intrusion, and 24% omission rates. FIRSSt attained 60% match, 15% intrusion, and 24% omission rates against 24hDR for all meals in the previous day. There was no evidence of sequence of assessment affecting accuracy indicators, but there was a weak effect of school lunch observation on percent intrusions. Obtaining a hair sample reduced the omission rate for FIRSSt vs 24hDR and increased the match rate for 24hDR vs observation, thereby enhancing this as a bogus pipeline procedure. Children generally enjoyed completing FIRSSt. Hispanic children were more likely to report problems using FIRSSt.
FIRSSt is somewhat less accurate than a dietitian-conducted 24hDR. However, this lower-cost procedure provides a promising method for assessing diet among children. Observation of consumption at school lunch may be reactive and artificially increase agreement. Obtaining a hair sample as a bogus pipeline may be a valuable technique for enhancing the accuracy of dietary assessment among children.