Reliability of 24-Hour Dietary Recall for Measuring Meal Timing
Although there is growing interest in evaluating the effects of meal timing on health parameters, limited dietary assessment methods capture meal-timing. Traditional food frequency questionnaires (FFQ) capture frequency of consumption of foods and beverages over time, but does not capture individual eating episodes or their timing. The 24-hour dietary recall captures timing of meal, snacks and beverages over the prior 24-hour period. Multiple 24-hour recalls can estimate habitual diet. The few prospective studies that include 24-hour recalls tend to do so for dietary calibration purposes, but often they are limited to a single 24-hour recall. Studies have investigated the reproducibility food, beverage and nutrient intake over time using 24-hour recalls, but none to our knowledge have measured the validity and reliability of measuring habitual meal timing patterns. It is therefore unknown whether a single 24-hour recall is adequate for meal timing measurement. This critical gap must be addressed prior to analyzing associations of meal timing parameters with disease outcomes in epidemiological studies, which we propose to do by leveraging data from the IDATA study.
We will evaluate the following meal timing variables from the IDATA ASA24.
(i) Duration of overnight fasting: the time between the last meal before the midpoint of time in bed and the first meal following the midpoint time in bed.
(ii) Midpoint of overnight fasting: midpoint of the overnight fasting period.
(iii) 24-hour distribution of caloric intakes: percent of total daily caloric intake in each of the six 4-hour periods spanning from midnight to midnight.
(iv) 24-hour distribution of macronutrient (carbohydrate, protein, and fat) intakes: percent of total macronutrient intake in each of the six 4-hour periods spanning from midnight to midnight.
(v) Ratio of later (15:00-24:00): earlier (06:00-14:00 ) energy intake.
Within and between-person variation in meal timing variables will be measured using the intraclass correlation coefficient (ICC) and Cohen’s kappa statistics. We will conduct mixed effects regression, adjusting for baseline covariates (e.g., age, physical activity, body mass index) to identify predictors of overnight fasting time, midpoint of overnight fasting, and distribution of caloric and macronutrient profiles. Predictors will include demographic (e.g., SES, income, race/ethnicity, age) and lifestyle (physical activity, smoking, alcohol use, dietary pattern).
Aim 1: Measure within and between-person variability of meal timing variables measured using 24-hour dietary recall.
Aim 2: Measure the correlation of a single 24-hour recall with the average of 2, 3, 4, 5, and 6 measures of meal timing variables.
Aim 3. Identify predictors of overnight fasting time, midpoint of overnight fasting, and distribution of caloric and macronutrient profiles.
Lacie Peterson, Qian Xiao, Tracy Layne, Cici Bauer, Ben Haaland, Mary Playdon