Comparison of eating timing metrics between a web-based dietary record and a dietitian-led interview

Background

Eating timing metrics are increasingly linked to chronic disease outcomes. However, the ability of web-based dietary records to capture these metrics in largescale studies has not been evaluated. Therefore, we aimed to assess agreement of eating timing metrics between the web-based 24-hr dietary record of the French NutriNet-Santé e-cohort, and an interview-based 24-hr recall.

Methods

In 2009, participants (N = 139, mean age=60.9±5.9y) completed a web-based record on a pre-arranged date, followed the next day by a dietitian-led phone interview using the same web interface to conduct a 24-hr recall. We computed the timing of the first and last eating occasions, the number of eating occasions, and the timing of the caloric midpoint. Agreement was assessed using the: (1) Wilcoxon signed-rank test, (2) intraclass correlation coefficients (ICC), and (3) Spearman’s rank correlation, in the full sample and stratified by sex, age, body mass index, educational level, and computer knowledge.

Results

Agreement between the two methods was high for the timing of the first and last eating occasions (Wilcoxon P ≥ 0.53; ICCs = 0.85-0.90; Spearman’s correlation coefficients = 0.86-0.95). For the number of eating occasions, agreement was moderate, but acceptable; fewer eating occasions were reported in the web-based record than in the interview-based recall (mean difference = -0.36; Wilcoxon P < 0.001), and the ICC and Spearman’s correlation coefficient were 0.72 and 0.70, respectively. Agreement for the caloric midpoint was similar (Wilcoxon P = 0.68; ICC = 0.65; Spearman’s correlation coefficient = 0.68). Stratified analyses showed higher agreements for the number of eating occasions and caloric midpoint among women and adults aged <60 years.

Conclusions

Eating timing metrics showed good to excellent agreement between a web-based 24-hr dietary record and an interview-based 24-hr recall. The NutriNet-Santé dietary records are an effective tool for assessing eating timing.

Key messages

• A web-based dietary record showed good agreement with an interview-based recall for eating timing metrics.

• Web-based 24-hr dietary records are effective tools to assess eating timing in largescale populations.