Assessing dietary reports in older adults: A comparative reliability study of different 24-h dietary recall (24HR)...

Technological advances have enabled the transition of dietary intake data collection, traditionally performed in person, to remote applications via telephone, mobile device, computer, or video calls. A notable example is the 24-Hour Dietary Recall (24HR), which can now be administered remotely through telephone or cellphone calls. This shift presents both benefits and drawbacks. While technological advancements have significantly enhanced the application of this instrument, the reliability of remotely administered 24HR, particularly those conducted via video call, remains underexplored in the current literature. Therefore, this study aimed to assess the reliability of 24HR data collected through telephone and video call interviews, comparing these results with those obtained from in-person administration.

We used data from two independent studies, referred to as Study A and Study B. Both databases comprise older adults (≥60 years old) from the Brazilian cities of Campinas, Limeira, Piracicaba, and the surrounding region. The researchers assessed food consumption through the administration of two 24HRs. The first one was conducted in-person - concomitantly with the collection of health data, socioeconomic conditions, and anthropometric measurements - while the second one was performed remotely (via telephone in Study A and video call in Study B). Statistical analysis employed a two-way mixed-effects model, absolute agreement for multiple measurements, using Intraclass Correlation Coefficients (ICC) with a 95% confidence interval. Additionally, Bland-Altman plots were employed to ascertain potential biases and limits of agreement. Linear mixed-effects models for repeated measures were performed for energy intake, considering the method (in-person X remote) as the tested term and the individual as the random-effects parameter, estimating crude models using only these two variables, and adjusted models including sex, age, and body mass index classification.

The findings revealed poor to moderate reliability for energy intake across both the telephone-administered 24HR (ICC values between 0.30 and 0.59) and video call modality (ICC values between 0.34 and 0.67). Similarly, the reliability estimates for protein, and fat intake ranged from poor to moderate, whereas carbohydrate intake demonstrated greater variability, with reliability ranging from poor to good. Overall, Study A (telephone) showed lower consistency across dietary components, while Study B (video call) presented a broader and generally higher range of reliability coefficients, reaching a good classification for carbohydrates in the group aged ≥75 years (ICC 0.78).

These data collectively suggest that the selection of the 24HR collection methodology may yield divergent nutritional estimates. This underscores the necessity for circumspection when choosing the approach for both population-based studies and clinical settings.