Impact of Early Time-Restricted Eating on Diet Quality, Meal Frequency, Appetite, and Eating Behaviors: A Randomized Trial

Abstract

Objective.

Time-restricted eating (TRE) can reduce body weight, but it is unclear how it influences dietary patterns and behavior. We therefore assessed the effects of TRE on diet quality, appetite, and several eating behaviors.

Methods.

Adults with obesity were randomized to early TRE plus energy restriction (eTRE+ER; 8-hour eating window from 07:00–15:00) or a control eating schedule plus energy restriction (CON+ER; ≥12-hour window) for 14 weeks. We assessed food intake via the Remote Food Photography Method, while eating patterns, appetite, and eating behaviors were assessed via questionnaires.

Results.

Fifty-nine participants completed the trial, of whom 45 had valid food records. eTRE+ER did not affect eating frequency, eating restraint, emotional eating, or the consistency of mealtimes relative to CON+ER. eTRE+ER also did not affect overall diet quality. The intensity and frequency of hunger and fullness were similar between groups, though participants in the eTRE+ER group were hungrier while fasting.

Conclusions.

When combined with a weight loss program, eTRE does not affect diet quality, meal frequency, eating restraint, emotional eating, or other eating behaviors relative to eating over a ≥12-hour window. Rather, participants implement eTRE as a simple timing rule by condensing their normal eating patterns into a smaller eating window.

INTRODUCTION

Time-restricted eating (TRE) is a form of intermittent fasting that limits eating to a consistent ≤10-hour daily window (1). Several studies report that TRE decreases body weight and improves cardiometabolic health (1, 2, 3, 4, 5). Although TRE can, in principle, be practiced with or without reducing calorie intake, several studies suggest when participants shorten their daily eating window, they spontaneously consume 10-25% fewer calories (6, 7, 8, 9, 10, 11, 12). Indeed, one cross-sectional study reported that each one-hour decrease in the daily eating duration was associated with eating 53 fewer kcal/day, and further, that TRE was associated with eating fewer calories, eating less often, and eating a lower glycemic load diet (13).

In addition to the daily eating duration, the timing of eating may also affect calorie intake and eating behavior. Cross-sectional analyses find that eating a greater proportion of calories in the evening and late-night eating are associated with greater energy intake, a higher body weight, impaired weight loss, and greater cardiometabolic risk (14, 15). Shifting a greater proportion of energy intake to the evening also blunts lipid oxidation (16, 17, 18) and impairs insulin sensitivity and hormone expression (19, 20). By contrast, interventions that shift food intake to the morning and/or eating earlier in the daytime consistently decrease appetite and/or food intake (21, 22, 23) and lead to greater weight loss (22, 24, 25, 26) relative to eating later in the day. In addition, diet quality tends to decline later in the day (27), whereas consuming meals early in the day improves diet quality (28, 29). Late night eating is also associated with greater energy intake (15), eating more often, and eating at more irregular times (30), which in turn have been linked to increased risk of cardiometabolic diseases (31). Thus, practicing TRE by eating earlier in the day (eTRE) could potentially reduce calorie intake; reduce snacking; limit the intake of calorie-dense, nutrient-sparse foods; and enforce more regular eating times.

Promoting a nutrient-dense, well-balanced diet with good eating habits is very important for health, as higher diet quality is associated with decreased risk for cardiovascular disease, cancer, and all-cause mortality (32). Yet, little is known about the effects of TRE on diet quality and food intake. Only three trials have assessed the effects of TRE on diet quality via the Healthy Eating Index (HEI), and they found no improvement in HEI scores compared to the control group (33, 34, 35). Though, one study found that the HEI score improved within the eTRE group but not the control group (35), while another study found that adherence to TRE reduced energy intake by decreasing carbohydrate and alcohol intake (36).

We recently conducted a moderately large weight-loss study comparing eTRE to an extended eating window of ≥12 hours. We previously reported that eTRE was superior to eating over a ≥12-hour period for losing body weight, resulting in 2.3 kg greater weight loss over the 14-week intervention (37). eTRE induced an additional energy deficit of 214 kcal/day as estimated using weight-loss modeling, but self-reported energy intake was not different between groups (37). eTRE also lowered diastolic blood pressure but did not affect other fasting cardiometabolic endpoints in the main intention-to-treat analysis. Herein, as tertiary outcomes of the parent study, we assessed the effects of eTRE on meal timing, eating frequency, diet quality, appetite, and eating behaviors. We hypothesized that eTRE would reduce energy intake relative to eating over a ≥12-hour period, whereas we had no a priori hypotheses on diet quality or eating behaviors.

METHODS

Participants.

This parallel-arm randomized controlled trial was approved by the Institutional Review Board at the University of Alabama at Birmingham (UAB; IRB number 300001207) and preregistered on ClinicalTrials.gov (NCT03459703). New patients of the UAB Weight Loss Medicine Clinic were enrolled between August 2018 and January 2020. Participants were eligible if they were 25-75 years of age, had a body mass index (BMI) of 30.0-60.0 kg/m², weighed less than 450 pounds (204.1 kg), and woke-up between 4:00-9:00 on most days. Participants were excluded if they had diabetes; were taking glucose-lowering or weight loss medications; lost or gained more than 5 pounds (2.3 kg) of weight in the past month; performed overnight shift work more than one day per week on average; regularly ate over a <10-hour daily period or ate dinner before 18:00; or had a major chronic condition that would affect patient safety or data validity (37).

Diet Protocols.

The study design and protocol are available in the primary outcome manuscript (37). In brief, participants were randomized either to eat within an 8-hour window between 7:00 and 15:00 (eTRE) or to eat over a self-selected ≥12-hour period (control, CON) for at least 6 days/week for 14 weeks. All participants followed the standard weight loss program prescribed by the UAB Weight Loss Medicine Clinic, which included regular weight checks, one-on-one dietary counseling with a registered dietitian to promote energy restriction (ER), recommendations to increase physical activity, and weekly group classes for support and accountability. In total, participants were provided with four one-on-one counseling sessions and instructed to attend at least 10 group classes. During the individual sessions, the dietitian instructed participants to reduce their energy intake by 500 kcal/day below their resting metabolic rate and then provided suggestions and examples customized to each participant’s usual diet to both reduce energy intake and eat healthier.

Outcomes.

We assessed several eating-related endpoints, including meal timing, meal frequency, food intake, diet quality, appetite, and eating behaviors. The timing of the eating window was assessed on a daily basis throughout the 14-week intervention. All other endpoints were assessed at baseline and at week 14. We measured food intake over a three-day period using the Remote Food Photography Method (RFPM), while the remaining eating-related endpoints were assessed via questionnaires that were administered in the morning following a ≥12-hour water-only fast.

Remote Food Photography Method (RFPM).

To assess food intake, we collected three-day digital food records using the Remote Food Photography Method© (RFPM) (38). Participants completed food records on two weekdays and one non-weekday. Participants used a smartphone app called SmartIntake^® to take “before” and “after” pictures of each item of food or beverage. To standardize the images, participants were instructed to take photos at a 45° angle at about an arm’s length away from the food and to place a fixed-sized reference card within the field of view. Participants recorded descriptions of what they consumed in the app and then categorized each eating episode as either breakfast, morning snack, lunch, afternoon snack, dinner, or evening snack, which allowed us to assess the distribution of food intake across the day. When food images were not captured or could not be transmitted, participants used a back-up method, such as a written food record or a verbal recall conducted over the phone. Using specialized software called the Food Photography Application©, trained dietitians identified a match for each food from the Food and Nutrient Database for Dietary Studies (FNDDS, version 6.0; USDA) and other sources, such as the manufacturer’s information and Nutrition Facts Panels. The dietitians also estimated the portion sizes using standard portion images and the reference card in the photo. Food and nutrient intakes were calculated as the difference between the selected food and plate waste. We considered three-day food records to be valid if participants recorded an estimated energy intake within 50% of predicted intake as estimated using the NIH Body Weight Planner, assuming a physical activity level (PAL) of 1.45. Data from the RFPM were used to estimate changes in eating frequency, the distribution of calorie intake across the day, energy intake, macronutrient composition, and diet quality (described below).

Diet Quality.

Data obtained from valid RFPM records were used to quantify diet quality using the Healthy Eating Index-2015 (HEI-2015) score. The HEI-2015 score (12) assesses conformance to the 2015 Dietary Guidelines for Americans (8), with higher scores indicating greater adherence to the guidelines. The 13 components in the 2015 iteration are adequacy of total fruits, whole fruits, total vegetables, greens and beans, whole grains, dairy, total protein foods, seafood and plant proteins, and fatty acids; and moderation of refined grains, sodium, added sugars, and saturated fats. Each of the components is scored on a density basis out of 1,000 calories, except for fatty acids, which is scored as the ratio of unsaturated to saturated fatty acids. As secondary measures of diet quality, we also assessed changes in fiber, cholesterol, fatty acids, the servings of key food groups, micronutrient intake, and alcohol and stimulant intake via the Munich Chronotype Questionnaire (MCTQ).

Adherence and Meal Timing.

Participants self-reported their meal-timing adherence by indicating when they started and stopped eating each day through electronic surveys automatically administered via REDCap software (39, 40). Participants were classified as adherent if they followed their assigned eating window within a ±30-minute grace period. Days with missing surveys were coded as non-adherent. We also assessed the standard deviation (SD) of the eating window as a measure of the regularity or consistency of mealtimes, with lower values indicating more consistent meal times.

Appetite.

To assess the frequency and intensity of hunger and satiety, we used the Appetite Questionnaire (AQ) and Retrospective Visual Analogue Scales (RVAS), and participants were asked to rate their appetite over the previous week. The AQ measures the frequency of various degrees of hunger and satiety via a 5-point Likert scale, ranging from 1 meaning “Never” to 5 meaning “Always.” For the RVAS, participants self-reported their average hunger, hunger during fasting, and fullness while eating on a 0-100 scale, with 0 meaning “Not at all” and 100 meaning “Extremely” (41).

Dutch Eating Behavior Questionnaire.

The Dutch Eating Behavior Questionnaire (DEBQ) assesses participants’ eating behaviors along three dimensions: emotional, external and restrained eating (42). Individual questions are scored on a 5-point Likert scale ranging from “Never” to “Very often,” and the composite scores for restrained, emotional, and external eating are tabulated, with higher numbers reflecting more of each trait.

Statistical Analyses.

Analyses were performed using two-sided tests with a type I error rate of α=0.05 in the software package R (version 4.0.3) (43). Analyses were performed in completers only. Continuous variables were analyzed using independent t-tests, while categorical data were analyzed using the chi-square test or Fisher’s exact test if the assumptions for the chi-square test did not hold. All results are expressed as mean ± SEM, except where otherwise indicated.

RESULTS

Participants.

We screened 656 people and enrolled 90 participants. Due to attrition (n=20) and inability to complete the protocol due to the COVID-19 pandemic (n=11), 59 participants completed all aspects of the intervention. Of the 59 completers, 45 participants had valid food records at both baseline and post-intervention. We included all 59 completers in the analyses of all endpoints, except for RFPM-derived endpoints, which were based on the 45 participants with valid food records. Participant characteristics are shown in Table 1. Total weight loss among completers was −4.3 kg (95% CI, −5.7 to −3.0 kg) in the CON+ER group and −6.6 kg (95% CI, −8.0 to −5.2 kg) in the eTRE+ER group, respectively (p=0.006).

Table 1.

Baseline Characteristics.

	Total (N=59)a	CON+ER (N=30)	eTRE+ER (N=29)
Characteristic	Mean ± SD	Mean ± SD	Mean ± SD	p
Demographics
Age, y	44 ± 11	43 ± 12	45 ± 10	0.69
Female, %	80%	80%	79%	1.00
Race, %				1.00
Black or African American	22%	23%	21%
Not Black or African American	78%	77%	79%
Ethnicity, %				0.25
Not Hispanic or Latino	92%	87%	97%
Hispanic or Latino	3%	3%	3%
Unknown or Not Reported	5%	10%	0%
BMI, kg/m2	39.1 ± 6.6	37.8 ± 5.5	40.5 ± 7.4	0.12
Eating Habits
Eating Duration, hours/day	12.8 ± 1.5	13.0 ± 1.7	12.6 ± 1.3	0.64
Eating Start Time, h:m	7:28 ± 1:04	7:32 ± 1:10	7:24 ± 0:59	0.16
Eating End Time, h:m	20:15 ± 1:26	20:31 ± 1:42	19:59 ± 1:04	0.33
Eating Episodes, number/day	3.9 ± 0.8	3.8 ± 0.8	3.9 ± 0.7	0.65
Meals	2.7 ± 0.3	2.6 ± 0.3	2.7 ± 0.3	0.31
Snacks	1.2 ± 0.7	1.2 ± 0.8	1.2 ± 0.6	0.97
Diet Quality (n=45)
Healthy Eating Index Score	51 ± 11	52 ± 13	51 ± 8	0.76
Fiber, g/day	14 ± 4	14 ± 5	13 ± 4	0.44
Added Sugar, g/day	14 ± 8	16 ± 9	12 ± 7	0.15
Cholesterol, mg/day	360 ± 166	341 ± 193	377 ± 139	0.47
Food Groups (n=45) a
Fruit, servings/day	0.5 ± 0.7	0.5 ± 0.8	0.5 ± 0.7	0.85
Vegetables, servings/day	1.5 ± 0.8	1.4 ± 0.6	1.6 ± 0.9	0.23
Legumes, servings/day	0.5 ± 0.8	0.7 ± 1.0	0.4 ± 0.6	0.25
Grains, servings/day	5.6 ± 2.0	5.9 ± 2.3	5.3 ± 1.7	0.37
Nuts and seeds, servings/day	0.7 ± 1.0	0.6 ± 0.9	0.7 ± 1.1	0.72
Eggs, servings/day	0.8 ± 0.8	0.7 ± 0.9	0.9 ± 0.6	0.63
Dairy, servings/day	1.4 ± 0.8	1.4 ± 0.8	1.4 ± 0.8	0.85
Fish, servings/day	0.8 ± 1.3	0.5 ± 1.2	1.1 ± 1.3	0.08
Meat, servings/day	4.6 ± 2.4	4.6 ± 2.5	4.7 ± 2.3	0.89
Eating Behavior
Dietary Restraint	14 ± 6	14 ± 5	14 ± 6	0.91
Emotional Eating	24 ± 13	26 ± 13	22 ± 12	0.32
External Eating	24 ± 6	24 ± 6	23 ± 6	0.30

Abbreviations: CON+ER, control eating schedule plus energy restriction; eTRE+ER, early time-restricted eating plus energy restriction; BMI, body mass index

^aFifty-nine of 90 participants randomized completed the trial and study questionnaires. Forty-five participants completed the trial and had valid digital food photography records (CON+ER, n=21; eTRE+ER, n=24).

^*p<0.05.

Meal Timing and Consistency.

Completers in the CON+ER group adhered to their assigned eating schedule 6.4 ± 0.8 days/week (mean ± SD), while those in the eTRE+ER group adhered 5.8 ± 0.8 days/week (mean ± SD). The most popular day that both groups chose to be non-adherent was Saturday, followed next by Friday and Sunday (data not shown). Figures 1A and 1B show the eating durations for the CON+ER and eTRE+ER groups, respectively. The CON+ER group ate over a 12.4 ± 0.2-hour window overall, while the eTRE+ER group ate within a 7.8 ± 0.1-hour window overall and a 7.0 ± 0.1-hour window on adherent days. On non-adherent days, the eTRE+ER group ate over a 10.4 ± 0.2-hour window. Figure 1C illustrates the participants’ eating windows. Both groups started eating at a similar time (p≥0.11) except on non-adherent days (p<0.001). However, the eTRE+ER group always stopped eating earlier than the CON+ER group (all p<0.001), even on non-adherent days, when they stopped eating about an hour earlier (p<0.001). We also assessed the consistency of eating times by calculating each individual’s SD and then averaging the SDs across all participants (Figures 1D–F). As shown in Figure 1D, both groups were similarly consistent in when they started eating (p≥0.29). However, the eTRE+ER group’s eating duration was more variable than the CON+ER group (1.0 ± 0.1 vs. 1.7 ± 0.1 hours; 0.7 ± 0.2 hours; p<0.001). This was due to greater variability in the eTRE+ER group’s eating duration on non-adherent days (1.2 ± 0.2 vs. 1.8 ± 0.1 hours; 0.6 ± 0.2 hours; p=0.006), as there was no difference on adherent days (0.7 ± 0.1 vs. 0.8 ± 0.1; 0.1 ± 0.1 hours; p=0.12).

Figure 1. Eating Windows.

(A) The CON+ER group ate over a 12.4-hour period on average but a 9.8-hour window when they were non-adherent. (B) The eTRE+ER group ate within a 7.8-hour window on average and a 7.0-hour window on adherent days but ate over a 10.4-hour window when they were not adherent. (C) Participants in both groups started eating at a similar time on average, weekdays, weekends, and adherent days. When they were not adherent, eTRE+ER participants started eating about 40 min later and stopped eating about 4 hours later, whereas CON+ER participants started eating nearly 2 hours later but otherwise stopped eating around their usual time. The eTRE+ER group stopped eating earlier than the CON+ER group in all cases. (D) Both groups were similarly consistent in when they started eating each day, as measured by the SD. (E) However, the eTRE+ER group was more consistent than the CON+ER group in their end time on adherent days but less consistent on non-adherent days. (F) The eTRE-ER group had a more variable eating duration due to non-adherent days. * p<0.05

Eating Frequency and Calorie Distribution.

eTRE+ER did not affect eating frequency. At week 14 (Figure 2A), the eTRE+ER group ate 3.5 ± 0.2 times/day, whereas the CON+ER group ate 3.7 ± 0.2 times/day (p=0.44), and there was no difference in the change scores between groups (−0.3 ± 0.3 times/day; p=0.19). eTRE+ER also did not affect the frequencies or sizes (kcal) of meals or snacks as separate categories (p≥0.10 for change scores; data not shown). However, eTRE+ER modestly affected the distribution of energy intake across meals and snacks. At week 14 (Figure 2B), the distribution of calorie intake across the day was mostly similar between groups, except the eTRE+ER group consumed a larger morning snack (4 ± 2% vs. 10 ± 2% of total kcal; p=0.02) and a smaller “dinner” (the last meal of the day) (36 ± 4% vs. 23 ± 3%; p=0.009) than the CON+ER group. However, the change scores were significantly different only for the mid-morning snack (6 ± 3%; p=0.045). On a cumulative basis, the eTRE+ER group ate 13 ± 5% more calories by the end of the morning snack (p=0.01), and their cumulative energy intake remained similarly higher until after dinner (data not shown).

Figure 2. Eating Frequency and Calorie Distribution.

(A) The frequency of meals and snacks did not change during the intervention and was not different between groups. (B) The distribution of calorie intake across the day at the end of the intervention was similar between groups, except the eTRE+ER group consumed 6% more of their total calories as a morning snack and 13% fewer of their total daily calories at “dinner” (the last meal of the day) relative to the CON+ER group. The changes in calorie distribution between baseline and week 14 were significant between groups only for the proportion of calories consumed at the morning snack. * p<0.05

Food Intake.

Food intake data are shown in Table 2 and Table 3. eTRE+ER did not affect energy intake and macronutrient composition, as previously reported (p≥0.53) (37). Neither intervention affected diet quality as assessed by the HEI score, and there were no differences between groups in the HEI score (−3.0 ± 5.1; p=0.56; Figure 3) or in other diet quality indicators, such as fiber, added sugar, cholesterol, saturated fat, other fatty acids, solid fat, or oils (p≥0.27). eTRE+ER also did not affect the intakes of various food groups, including grains, fruit, vegetables, legumes, nuts and seeds, eggs, meat, dairy, fish, and alcohol relative to CON+ER (p≥0.07). The one exception is that eTRE+ER was less effective at reducing refined grain intake as measured by the HEI component score (−2.5 ± 1.2; p=0.03). There were no differences in any of the other 12 HEI component scores (p≥0.12), Similarly, there were no differences in the intakes of most micronutrients, except for copper and vitamin D, which decreased in the eTRE+ER group (Table 3).

Table 2.

Changes in Food Intake and Diet Quality.

	CON+ER Within-Group Change		eTRE+ER Within-Group Change		Between-Group Difference
Dietary Endpoint	Mean ± SEM	p	Mean ± SEM	p	Mean ± SEM	p
Diet Quality (n=45) a
Healthy Eating Index Score	7.5 ± 3.6	0.052	4.5 ± 3.5	0.22	−3.0 ± 5.1	0.56
Fiber, g/day	0 ± 1	0.83	−1 ± 1	0.63	−1 ± 2	0.64
Added Sugar, g/day	−9 ± 2	<0.001*	−6 ± 2	0.008*	3 ± 3	0.27
Solid Fat, g/day	−12 ± 2	<0.001*	−16 ± 3	<0.001*	−4 ± 4	0.30
Cholesterol, mg/day	−70 ± 40	0.10	−70 ± 39	0.08	0 ± 56	1.00
Oils, g/day	−12 ± 4	0.008*	−11 ± 3	<0.001*	1 ± 5	0.81
Monounsaturated Fat, g/day	−9 ± 2	<0.001*	−10 ± 2	<0.001*	−1 ± 3	0.71
Polyunsaturated Fat, g/day	−7 ± 2	0.007*	−7 ± 1	<0.001*	0 ± 3	0.92
Omega-3 fatty acids, g/day	−1 ± 0	0.053	0 ± 0	0.04*	0 ± 0	0.58
Omega-6 fatty acids, g/day	−6 ± 2	0.01*	−6 ± 1	<0.001*	0 ± 2	0.95
Food Groups (n=45) a
Fruit, servings/day	0.2 ± 0.2	0.28	0.2 ± 0.2	0.18	0.0 ± 0.3	0.97
Vegetables, servings/day	0.3 ± 0.2	0.29	−0.4 ± 0.3	0.15	−0.7 ± 0.4	0.07
Legumes, servings/day	−0.3 ± 0.2	0.29	0.0 ± 0.2	0.94	0.3 ± 0.3	0.38
Grains, servings/day	−2.5 ± 0.5	<0.001*	−1.6 ± 0.6	0.007*	0.8 ± 0.7	0.28
Nuts and seeds, servings/day	−0.1 ± 0.2	0.71	−0.3 ± 0.3	0.28	−0.2 ± 0.4	0.53
Eggs, servings/day	−0.1 ± 0.2	0.68	0.0 ± 0.2	0.81	0.0 ± 0.3	0.87
Dairy, servings/day	−0.2 ± 0.2	0.24	−0.4 ± 0.1	0.01*	−0.2 ± 0.2	0.39
Fish, servings/day	−0.1 ± 0.3	0.63	−0.6 ± 0.3	0.04*	−0.5 ± 0.4	0.27
Meat, servings/day	−0.6 ± 0.7	0.43	−0.3 ± 0.4	0.56	0.3 ± 0.8	0.70
Stimulants (n=59 except where otherwise noted) a
Alcohol, g/day (n=45)	0 ± 0	0.41	−3 ± 1	0.03*	−2 ± 1	0.09
Beer, servings/wk	−0.1 ± 0.1	0.06	−0.1 ± 0.1	0.41	0.0 ± 0.1	0.81
Wine, servings/wk	0.0 ± 0.2	0.96	−0.1 ± 0.1	0.46	−0.1 ± 0.2	0.67
Liquor, servings/wk	0.0 ± 0.1	0.63	0.1 ± 0.1	0.45	0.1 ± 0.1	0.38
Caffeine, mg/day (n=45)	−57 ± 16	0.002*	−47 ± 17	0.01*	10 ± 24	0.68
Coffee, servings/wk	−1.7 ± 0.7	0.02*	−3.3 ± 1.7	0.06	−1.6 ± 1.8	0.37
Tea, servings/wk	−0.9 ± 0.9	0.30	0.0 ± 1.2	0.99	0.9 ± 1.5	0.54
Caffeinated Soda, servings/wk	−0.6 ± 0.7	0.43	−0.3 ± 0.6	0.60	0.2 ± 1.0	0.80

Abbreviations: CON+ER, control eating schedule plus energy restriction; eTRE+ER, early time-restricted eating plus energy restriction.

^aFifty-nine of 90 participants randomized completed the trial and study questionnaires. Forty-five participants completed the trial and had valid digital food photography records (CON+ER, n=21; eTRE+ER, n=24).

^*p<0.05.

Table 3.

Changes in Micronutrient Intake.

	CON+ER Within-Group Change		eTRE+ER Within-Group Change		Between-Group Difference
Micronutrients (n=45)a	Mean ± SEM	p	Mean ± SEM	p	Mean ± SEM	p
Calcium, mg/d	−103 ± 50	0.051	−171 ± 78	0.04*	−68 ± 96	0.48
Iron, mg/d	−2.5 ± 1.0	0.02*	−3.0 ± 1.1	0.01*	−0.5 ± 1.5	0.75
Magnesium, mg/d	−1 ± 20	0.98	−35 ± 24	0.16	−34 ± 32	0.29
Phosphorus, mg/d	−190 ± 77	0.02*	−261 ± 76	0.002*	−71 ± 108	0.52
Potassium, mg/d	−29 ± 175	0.87	−480 ± 148	0.004*	−451 ± 227	0.053
Sodium, mg/d	−941 ± 208	<0.001*	−733 ± 202	0.001*	208 ± 291	0.48
Zinc, mg/d	−0.8 ± 0.6	0.22	−2.1 ± 1.0	0.04*	−1.4 ± 1.2	0.26
Copper, mcg/d	0.0 ± 0.1	0.85	−0.3 ± 0.1	0.006*	−0.3 ± 0.1	0.03*
Selenium, mcg/d	−23 ± 9	0.02*	−25 ± 10	0.01*	−3 ± 13	0.85
Vitamin A Equivalents, mcg/d	214 ± 93	0.03*	38 ± 132	0.78	−176 ± 166	0.29
Vitamin D, mcg/d	1 ± 1	0.26	−1 ± 1	0.08	−2 ± 1	0.04*
Vitamin C, mg/d	26 ± 13	0.06	6 ± 19	0.74	−20 ± 24	0.42
Thiamin, mg/d	−0.3 ± 0.1	0.01*	−0.3 ± 0.1	0.04*	0.1 ± 0.2	0.66
Riboflavin, mg/d	−0.3 ± 0.1	0.02*	−0.2 ± 0.1	0.14	0.1 ± 0.2	0.52
Niacin, mg/d	−4.5 ± 2.0	0.03*	−3.5 ± 2.0	0.11	1.1 ± 2.9	0.71
Vitamin B6, mg/d	0.1 ± 0.2	0.51	−0.2 ± 0.2	0.47	−0.3 ± 0.3	0.33
Folate, mcg/d	−99 ± 41	0.03*	−34 ± 42	0.42	65 ± 59	0.27
Vitamin B12, mcg/d	−0.7 ± 0.5	0.13	−1.0 ± 0.5	0.055	−0.3 ± 0.7	0.65
Vitamin K, mcg/d	−12 ± 23	0.59	53 ± 37	0.17	65 ± 45	0.16
Vitamin E, mg/d	1.1 ± 1.7	0.52	0.5 ± 1.9	0.79	−0.6 ± 2.6	0.82

Abbreviations: CON+ER, control eating schedule plus energy restriction; eTRE+ER, early time-restricted eating plus energy restriction.

^aFifty-nine of 90 participants randomized completed the trial and study questionnaires. Forty-five participants completed the trial and had valid digital food photography records (CON+ER, n=21; eTRE+ER, n=24).

^*p<0.05.

Figure 3. Diet Quality.

(A) Diet quality as assessed by the Healthy Eating Index-2015 (HEI) did not change in either group during the intervention, nor were there any between-group differences. Similarly, other common indicators of diet quality—such as (B) fiber and (C) added sugar—were similar between groups.

Hunger and Fullness.

Appetite data are shown in Figure 4. eTRE+ER did not affect the frequencies of various degrees of hunger and fullness as measured by the AQ (all p≥0.11; Figure 4A) or average hunger as measured by RVAS (−7 ± 6 mm; p=0.24; Figure 4B). CON+ER was more effective at suppressing hunger while fasting (24 ± 9 mm; p=0.008; Figure 4C), while the two interventions led to similar fullness or satiation while eating (7 ± 8 mm; p=0.38; Figure 4D).

Figure 4. Hunger and Fullness.

(A) There were no differences in the frequencies of various degrees of hunger and fullness between groups, as assessed by the Appetite Questionnaire. Ratings of hunger assessed by retrospective visual analog scales indicated that (B) average hunger and (D) fullness while eating were similar between groups. However, (C) eTRE+ER was less effective at suppressing hunger during the fasting period than CON+ER.

* p<0.05

Eating Behavior.

Data from the Dutch Eating Behavior Questionnaire are shown in Figure 5. eTRE+ER did not affect eating restraint (−1 ± 2; p=0.42), emotional eating (1 ± 2; p=0.49), or external eating (1 ± 1; p=0.47).

Figure 5. Eating Behavior.

There were no between-group differences in (A) eating restraint, (B) emotional eating, or (C) external eating, as assessed by the Dutch Eating Behavior Questionnaire.

DISCUSSION

We recently conducted one of the first relatively large weight-loss studies to compare TRE versus eating over a ≥12-hour period and found that eTRE was superior for losing weight (37). Here, we investigated the effects of eTRE on several dietary outcomes, including meal timing, eating frequency, diet quality, appetite, and eating behaviors.

We thoroughly investigated how participants adopted their assigned eating windows in the real-world. The eTRE+ER group ate over a 7.0-hour window ending a few minutes before 15:00 when they adhered but ate over a 10.4-hour window when they were non-adherent. Contrary to our expectations, the eTRE+ER group did not eat over a ≥12-hour period on non-adherent days and still ate over a shorter eating window than at baseline. The most popular day both groups chose to be non-adherent was Saturdays, followed next by Fridays and Sundays. We also examined how consistent participants were with the timing and duration of their eating windows. On adherent days, the eTRE+ER group was very consistent at stopping eating around 15:00. However, on non-adherent days, they were less consistent, resulting in the eTRE+ER group having a more variable eating duration overall. Surprisingly, we found that eTRE does not improve the regularity of the eating window, and may instead increase the variability of the eating window due to participants being non-adherent and/or having break days.

In our study, we also found that eTRE did not reduce the number of meals or snacks, which is in contrast to at least one previous study (44). This null finding may stem from the weight loss programming in this trial. In addition to providing counseling to restrict energy intake, a registered dietitian encouraged participants in the eTRE+ER group to eat their usual number of meals and snacks within a shorter eating window so that they would ingest a sufficient amount of food before the end of their eating window. Thus, our findings may not generalize to people who are following eTRE without the support of a dietitian or other interventionist. We also found that the eTRE+ER group did not majorly change the sizes of various meals and snacks relative to the CON+ER group. The only significant difference is that the eTRE+ER group ate a modestly larger mid-morning snack than the CON+ER group and also modestly increased the cumulative proportion of calories they ate for breakfast and the mid-morning snack combined. Taken together, this suggests that our participants practiced eTRE primarily by condensing the same number of meals and snacks into a shorter eating window, rather than by eating less often or markedly changing the sizes of their meals or snacks.

Eating earlier in the day has also been linked to better diet quality (27). However, in our trial, eTRE+ER did not affect diet quality as assessed by the HEI score either relative to baseline or the CON+ER group. Our findings are consistent with previous research suggesting that TRE does not influence diet quality (33, 34, 35). Interestingly, eTRE+ER did not affect alcohol consumption and was less effective in reducing refined grains, which could stem from the eTRE+ER group relying more on processed breakfast foods such as cereal, bread, and baked goods and/or the need to have readily available snack foods. Maleab et al. (7) did find that TRE reduces snacking and specifically reduced the intake of high-quality snacks. We did not explore whether eTRE affected food choices at different times of the day. Overall, there were no major or overarching thematic differences in food or nutrient intake either relative to baseline or between groups. This may be due to our limited sample of completers with valid diet records. Our largely null results could also be because participants in both groups received the same dietary counseling promoting healthier food options, which may have masked the effects of eTRE alone. Therefore, our results may not generalize to people who follow TRE outside the context of a weight management program or a research study. Nonetheless, our data suggest that eTRE does not meaningfully affect food intake, beyond potentially decreasing energy intake.

We also investigated the effects of eTRE on self-reported appetite. eTRE did not affect the frequency or intensity of hunger or satiety, though it did increase feelings of hunger during the fasting period. This conflicts with other studies reporting that TRE reduces appetite and improves appetite and satiety hormones (1, 4, 10, 45). Since some of these studies matched calorie intake across groups and/or were conducted in energy balance, one possible way to reconcile these findings is that TRE could enable people to sustain a larger calorie deficit at the same level of hunger; this deserves further exploration. It has also been suggested that intermittent fasting interventions, such as TRE, induce weight loss by increasing restraint and decreasing emotional eating, thereby curbing snacking. However, eTRE did not affect dietary restraint, emotional eating, or external eating as assessed by the DEBQ, nor did it affect snacking in our study. One criticism of intermittent fasting is that TRE may perpetuate emotional eating and negatively impact eating behaviors, such as binge eating. Similar to Gabel et al. (46), we did not observe detrimental effects of TRE on eating behavior. However, this area warrants further investigation as individuals with disordered eating tendencies are often excluded from weight loss trials.

Our study has both strengths and limitations. Strengths include that we have conducted one of the largest and most detailed investigations into eating behaviors and diet quality, with a variety of assessments. Limitations include a reduced sample size due to attrition and the COVID-19 pandemic and that we had to exclude some food records due to severe underreporting. As a result, we were only powered to detect rather large effects in eating behaviors. In addition, all eating-related endpoints were analyzed only in completers and were collected via self-report, which may be impacted by response bias. This includes the retrospective appetite assessments, which can also be subject to recall bias. Appetite is more accurately assessed using standardized meal tests. Further, we did not collect dietary data at multiple time points and only collected dietary data at baseline and during the last week of the intervention, which may not adequately represent how participants ate during the course of the study. We also primarily tracked meal timing adherence and not adherence to the energy restriction recommendation—a decision we made to limit participant burden. Lastly, we cannot rule out sampling bias, namely, that individuals who were more interested in or could more readily adopt eTRE enrolled in our trial.

We conclude that at least when combined with a weight loss program, eTRE does not appreciably affect eating frequency, meal and snack sizes, overall diet quality, eating restraint, emotional eating, and other eating behaviors relative to eating over a ≥12-hour period. eTRE slightly changed the distribution of energy intake across meals and snacks. Contrary to popular claims, eTRE also did not improve the consistency of the eating window, due to the large increase in the eating duration on non-adherent days; though, participants did eat at more consistent times on adherent days. The consequence of this increased variability in the eating window due to break days and whether it has any negative metabolic or circadian consequences merits further investigation. Importantly, eTRE did not increase overall hunger despite producing greater weight loss, though hunger was higher while fasting. Rather, we conclude that individuals following eTRE incorporate it primarily as a timing rule and condense their habitual eating habits into a smaller window, rather than changing the number, size, or food content of meals and snacks. Therefore, TRE is a simple, low-structure dietary intervention which allows increased flexibility in eating choices by focusing solely on meal timing.

Study Importance Questions.

What is already known about this subject?

• By shortening the daily eating duration, time-restricted eating (TRE) can spontaneously reduce energy intake and induce modest weight loss.

• The benefits of TRE may depend on the timing of food intake as well as the eating duration.

What are the new findings in your manuscript?

• When combined with a weight management program, practicing TRE by eating early in the day (eTRE) did not appreciably affect diet quality, appetite, eating restraint, emotional eating, or other eating behaviors.

• Participants adopted eTRE by slightly shifting a greater proportion of their food intake to the morning but did not change the frequency or relative sizes of meals and snacks.

How might your results change the direction of research or the focus of clinical practice?

• Participants implement TRE primarily as a timing rule and do not make significant changes to their meal frequency, diet quality, or eating behaviors.

• Contrary to expectations, TRE does not improve diet quality, eating restraint, or emotional eating or reduce eating at irregular times—at least not in the context of a weight loss program.