Four-week time restricted eating intervention is associated with improvements in cardiometabolic risk factors and dysregulated eating among emerging adult women: a single-armed trial

Abstract

Purpose

Time restricted eating (TRE) improves cardiometabolic (CM) health. However, TRE’s usefulness as a preventative intervention, particularly among women at risk for dysregulated eating is unknown. This single-arm study examined the impact of a 4-week TRE intervention on eating behaviors, body composition, and dietary intake in women at risk for dysregulated eating.

Methods

36 emerging adult women with eating windows ≥ 12 h and moderate–high dietary restraint completed 1 baseline week and 4 weeks of TRE (10 h eating window ending by 8 pm). Participants completed the Dutch Eating Behavior Questionnaire, 3-day food logs, anthropometric measurements, and DXA scans at baseline and post-intervention. Ecological momentary assessments (EMA) were administered 5x/day during baseline and weeks 1 and 4 of TRE to assess eating in the absence of hunger (EAH).

Results

Adherence with TRE and EMA was above 85%. Significant decreases occurred in emotional eating (p = 0.009, Cohen’s d = − 0.470), caloric intake (p < 0.001, Cohen’s d = − 0.750), body weight (p < 0.001, Cohen’s d = − 0.649), and visceral fat (p = 0.026, Cohen’s d = − 0.452) between baseline and end of TRE. The odds of EAH were significantly lower during TRE than baseline (week 1 OR: 0.48, 95% CI 0.32–0.71; week 4 OR: 0.46, 95% CI 0.29–0.71, both p < 0.001). Compared to baseline, EAH decreased during fasting and the first 4 h of the eating window (p = 0.002–0.033), but not later in the eating window (p = 0.579–0.763).

Conclusions

A brief TRE intervention reduced emotional eating and visceral fat in emerging adult women. However, results are preliminary and caution is still warranted when implementing restrictive interventions in this population.

Clinicaltrials.gov identifier: NCT06145009, submitted 10/17/2023.

Level III: Evidence obtained from well-designed cohort or case–control analytic studies.

Introduction

Time restricted eating (TRE) may be a promising strategy for cardiovascular disease (CVD) prevention if TRE does not inadvertently increase dysregulated eating. TRE is the practice of limiting daily eating to a specified time range, typically 4–10 h in length, without requiring restriction of energy intake or types of foods consumed [1]. TRE interventions reduce cardiometabolic risk factors such as elevated blood glucose, triglycerides, and blood pressure [2–4], thereby reducing risk for CVD [5]. Exposure to risk factors throughout the lifespan drives CVD development, and CVD risk factors are present early in life [6]. Over half of emerging adults (age 18–29 years) have at least one risk factor for metabolic syndrome [7, 8] and atherosclerotic plaques are present in 28% of young adults [9]. Emerging adults commonly have long eating windows and eat late in the evening [10], making TRE a potentially valuable intervention. However, emerging adults, particularly emerging adult women, are vulnerable to dysregulated eating [11–15] and the impact of TRE interventions on dysregulated eating is not yet clear [16].

Dysregulated eating, ranging from eating in the absence of hunger, emotional eating, and night eating to loss of control eating and binge eating, increases cardiometabolic risk [17]. Dysregulated eating is common among restrained eaters, who believe they should limit food intake for weight loss or management. Restrained eaters are vulnerable to disinhibition (eating in response to external cues or emotions) when their dietary rules are compromised [18, 19]. Disinhibition is problematic as it has been associated with elevated cardiometabolic risk factors, including higher weight [20, 21], body mass index (BMI) [20, 21], blood pressure [22], blood lipids [22], and hemoglobin A1c [23]. Similarly, emotional eating has been associated with greater odds of excess weight, hypertension, type 2 diabetes, and increased cardiovascular risk over time [24–26]. Eating in the absence of hunger (EAH) is associated with lower diet quality, weight gain, and higher levels of dietary restraint, loss of control eating, and hedonic hunger [21, 22, 27, 28]. Progression of dysregulated eating behaviors to disordered eating or eating disorders would further increase cardiovascular risk, as eating disorders negatively impact the cardiovascular system [17, 29].

Limited and mixed evidence complicates our understanding of the impact of TRE interventions on dysregulated eating behaviors. Cross-sectional studies have found that intermittent fasting, including TRE, is associated with binge eating in young adults [30, 31]. In contrast, in a study of 10 dancers who underwent a TRE intervention, no change in disordered eating was observed [32]. A systematic review found that in three qualitative studies of primarily middle-aged adults, nighttime snacking and mindless eating decreased with TRE interventions, but EAH occurred during the eating window [16]. As EAH varies throughout the day and TRE may impact EAH differently at different times of the day, it is possible that a TRE intervention may limit EAH or that EAH may simply shift to a different time of day. To our knowledge, no studies have quantitatively examined within-day changes in EAH throughout a TRE intervention.

Taken together, it is critical that the impact of TRE interventions on emerging adult women be clarified in terms of whether they exacerbate dysregulated eating, potentially negating cardiometabolic benefit. As such, the present study examined the extent to which a 4-week TRE intervention impacted eating behavior and associated cardiometabolic risk factors among emerging adult women. We hypothesized that the eating behaviors of emotional eating, external eating, and EAH would decrease from baseline to end of treatment. We also expected that there would be changes in cardiometabolic risk factors across the intervention, including improved diet quality and decreased calorie intake, body weight, percent body fat, and visceral fat mass. We further hypothesized that there would be within-day changes in EAH, with decreased EAH at night and increased EAH during the end of the eating window. As the intervention did not require changing the types or amounts of food consumed, we predicted that dietary restraint would remain unchanged.

Methods

Participants

Participants were recruited from a large mid-Atlantic university and the surrounding community via flyers distributed around campus and local businesses, disseminated in classes and student groups, and posted on social media. Women aged 20–29 years were included if they could speak and read English, owned a smart phone, and had a BMI ≥ 20 kg/m² (due to potential weight loss), a typical eating window of ≥ 12 h with their last eating occasion typically occurring after 8 pm, and a moderate–high baseline level of dietary restraint. Those aged 18–19 were excluded to prevent potential confounding with dietary changes that can occur when transitioning to college [33]. Dietary restraint was assessed with the question: “Please respond to this statement on a scale of 1–9, with 1 meaning never and 9 meaning always: I try to limit the type or amount of foods I consume in order to control my body weight or shape.” Those who answered 4 or higher were eligible. Exclusion criteria were current pregnancy, nursing, or planning to become pregnant, working night shifts, a past or current eating disorder diagnosis, or having a chronic medical condition or condition that requires a special diet or specified meal timing. A power analysis indicated a sample size of 36 participants would provide 80% power to detect a large (f = 0.40) effect size at an alpha of 0.05 for each of the primary eating behavior outcomes as measured by the Dutch Eating Behavior Questionnaire (DEBQ). Large effect sizes have been seen in the DEBQ subscale scores with nutrition interventions [34, 35], and represent a clinically significant change. The study protocol was preregistered on clinicaltrials.gov (NCT06145009) and approved by the University of Delaware’s IRB. All participants provided written informed consent prior to participating.

Study Design

This non-controlled trial utilized a single arm design. One week of baseline data were collected followed by a 4-week TRE intervention. Participants attended in-person study visits at baseline and the end of TRE, completed 3-day food logs during baseline and week 4 of TRE, and wore an actigraph and completed ecological momentary assessment (EMA) measures during baseline, week 1 of TRE, and week 4 of TRE (see Fig. 1).

Fig. 1.

Diagram of study procedures

Intervention

During the TRE intervention participants were instructed to consume all food and caloric beverages within a 10-h eating window that ended by 8 pm, based on findings that suggest 8 pm is a beneficial cutpoint for favorable body composition changes in this population [36]. Participants self-selected their eating window within these parameters. Participants were informed that they did not need to make any changes to the types or amounts of foods consumed. Outside of the eating window, participants were encouraged to consume ample fluids: water, black coffee, brewed tea, flavored water, and other non-caloric beverages were permitted.

Measures

Demographics

At baseline, participants self-reported their age, race, ethnicity, and education level.

Outcome measures: Eating behaviors, anthropometrics, and body composition measures were completed at the baseline and end of TRE study visits.

Eating behaviors

Eating behaviors were assessed with the Dutch Eating Behavior Questionnaire (DEBQ), a 33-item survey assessing dietary restraint, emotional eating, and external eating [37]. All items are scored on a Likert scale from 1 (never) to 5 (very often). Sample questions include: Do you try to eat less at mealtimes than you would like to eat? (restraint) Do you have a desire to eat when you are feeling lonely? (emotional), and If you see others eating, do you also have the desire to eat? (external) [37]. Subscale total scores are divided by the number of items on that scale to produce a standardized score from 1 to 5 [37]. All subscales showed acceptable internal consistency (Cronbach’s alpha: 0.822 for restraint, 0.935 for emotional eating, and 0.782 for external eating).

Anthropometrics

Height was measured to the nearest 0.1 cm using a Holtain Harpenden wall-mounted stadiometer. Participants were positioned with feet at 60 degrees and heels, buttocks, shoulders, and the back of the head touching the device and the head positioned in the Frankfurt plane, with measurement taken at the end of an inhalation. Weight was measured to the nearest 0.1 kg using a Scale-Tronix 5002 Bariatric Stand-on digital floor scale, with the participant in light clothes and shoes removed. Measurements were repeated three times and averaged. BMI was calculated by dividing weight in kilograms by height in meters squared [38].

Body composition

Body composition was measured with a Hologic Horizon-A Dual Energy X-ray Absorptiometry (DXA) machine (software version 5.6.0.4). Participants were instructed to fast for 2 h prior to their scan. Prior to scanning, participants removed all metal and jewelry and completed a pre-test questionnaire, including questions confirming the participants were not pregnant. Whole body measurements for lean and fat soft tissue were conducted by a trained and licensed DXA operator. Participants were scanned two consecutive times at visit 2 and the coefficient of variation was calculated for body fat percentage to determine precision.

Eating in the absence of hunger and compliance

EAH and compliance were assessed using ecological momentary assessment (EMA) during the baseline week and weeks 1 and 4 of TRE. EMA allows real-time information to be captured about individuals in their natural environment by prompting them to fill out surveys on their mobile device [39]. Participants were prompted via text message to complete EMA surveys 5 times per day using the MyCap application [40]. Participants downloaded MyCap to their phone during their baseline visit and received training on how to fill out the surveys. Interval-contingent prompts were sent at the times corresponding to the start and end of each participant’s eating window and the participant’s usual bedtime, and signal-contingent prompts were sent once in the first 4 h of the eating window and once in the second 4 h of the eating window (see Table 1).

Table 1.

Baseline participant characteristics

Eating in the absence of hunger was defined as an EMA survey in which a participant reported that they had eaten but did not choose “I was a little hungry” or “I was very hungry” as a reason for eating.

Dietary intake

Participants completed a 3-day food record using the Meal Logger app during their baseline period and week 4 of TRE. Meal Logger allows participants to submit time-stamped photos of the foods and drinks consumed, as well as to provide descriptions of food items [41]. At the baseline visit, participants received instructions on recording food intake. A research assistant reviewed the logs and asked any needed clarifying questions through the app. Diet data were analyzed using Nutrition Data System for Research software version 2023, developed by the Nutrition Coordinating Center (NCC), University of Minnesota, Minneapolis, MN [42]. Total calories and Healthy Eating Index-2015 (HEI) scores were calculated by NDSR. HEI assesses adherence to the Dietary Guidelines for Americans, including adequate consumption of fruits, vegetables, whole grains, protein and dairy, and limited intake of sodium, added sugar, and saturated fat [43]. HEI scores range from 0 to 100, which higher scores indicating higher diet quality.

Activity and sleep

The Actigraph GT3X was worn on participants’ non-dominant wrist during the baseline week and weeks 1 and 4 of TRE to estimate time spent in activity and sleep. Participants were instructed to wear the actigraph at all times unless the device might get wet (bathing, swimming, etc.). Actigraph data were processed with ActiLife software using 60 s epochs for days with at least 16 h of wear time. Sleep times that exceeded three standard deviations from the population mean were excluded [44]. Mean daily minutes of moderate to vigorous physical activity (MVPA) and sleep duration were calculated and reported descriptively, as these factors can also influence eating and body weight.

Statistical analysis

Data were analyzed using SPSS Version 29 (SPSS Inc., Chicago, IL, USA). Descriptive data were visually assessed for normality and reported as mean (standard deviation), median (interquartile range) or n (%). Differences between participants who completed the TRE intervention and those who did not were assessed with independent samples t tests, Mann–Whitney U test or chi-square tests.

Adherence to EMA was assessed by calculating the percentage of EMA surveys participants completed during the baseline week and weeks 1 and 4 of TRE. Adherence to TRE was assessed using the first and last EMA surveys of the day. EMA survey 1 was sent at the start of the eating window and asked if participants had consumed food or non-caloric beverages since bedtime the night prior, and EMA survey 5 was sent at bedtime and asked if participants had consumed food or non-caloric beverages since their eating window closed. If participants endorsed eating on EMA survey 1 or 5, indicating that they had eaten during their fasting window, they were considered non-adherent for that day.

To ascertain changes in DEBQ subscales for emotional eating and external eating from baseline to end of TRE, one-sided paired t tests were used. A series of three generalized linear mixed models (GLMM) assessed EAH. All models used a binomial distribution and logit link function with a compound symmetry covariance structure, as this structure provided the best model fit. The first model (Model 1) addressed the aim of assessing change in EAH between baseline and TRE. During EMA surveys in which participants reported eating, the possible outcomes were eating with hunger or EAH. The model was fitted with the presence of EAH during eating occasions as a binary outcome, week of study as a fixed factor, a random intercept for participant, and a random slope for participants’ individual change over time. β coefficients reported as odd ratios with 95% confidence intervals and predicted probabilities at each timepoint are reported.

To estimate the within-day changes in EAH, two GLMMs were used. First, a model (Model 2) assessed the presence of EAH during eating occasions. The model was fitted with week of study, EMA survey number, and the interaction of week of study by EMA survey number as fixed factors, a random intercept for participant, and a random slope for participants’ week of study by EMA survey number interaction. The model was then repeated with the outcome of presence of EAH during all EMA surveys (i.e., when EAH occurred compared to when eating with hunger occurred and when eating did not occur, Model 3). Significant interaction effects were further explored with simple contrasts comparing differences in EAH at each EMA survey timepoint between baseline and week 1 of TRE and baseline and week 4 of TRE, with a Sidak adjustment for multiple comparisons.

Changes in HEI-2015 score, total calories, body weight, percent body fat, and visceral fat mass from baseline to end of TRE were assessed using one-sided paired t tests. Differences in mean minutes of MVPA and sleep time between baseline and end of TRE were assessed with two-sided paired t tests. Effect sizes for all paired t tests were assessed with Cohen’s d, with 0.2 considered a small effect, 0.5 considered a medium effect, and 0.8 considered a large effect [45]. Significance was set at alpha of 0.05 for all tests.

Results

Participant characteristics

Between September 2023 and April 2025, 616 individuals were screened for study inclusion and 118 met study inclusion criteria. Of the 118 eligible, 36 were enrolled between October 2023 and April 2025, and 30 completed the study (see Fig. 2 for participant flow). The most common reason for withdrawing was being too busy to complete the study. Demographic characteristics are shown in Table 2. There were no significant differences in baseline demographic measures between those who did and did not complete the study. Completers had a median age of 22 years. The majority were White (66.7%) college undergraduates (56.6%). Participants reported 86.0% adherence to TRE.

Fig. 2.

Participant flow diagram

Table 2.

Baseline participant characteristics

	Completers n = 30	Non-completers n = 6	p
Age (years, Median (IQR))	22.0 (20.75–24.25)	22.0 (20.75–27.5)	0.621
Hispanic/Latino (yes, n (%))	1 (3.3%)	0 (0%)	0.650
Race (n, %)
American Indian/Alaska Native	0 (0%)	0 (0%)	–
Asian	9 (30%)	1 (16.7%)	0.056
Black or African American	3 (10%)	0 (0%)	0.418
Native Hawaiian/Pacific Islander	0 (0%)	0 (0%)	–
White	20 (66.7%)	5 (83.3%)	0.418
Other	0 (0%)	0 (0%)	–
Current education level (n, %)			0.589
College—second year	4 (13.3%)	0 (0%)
College—third year	6 (20%)	2 (33.3%)
College—fourth year	7 (23.3%)	0 (0%)
Completed Undergraduate Degree	1 (3.3%)	0 (0%)
In Graduate School	8 (26.7%)	3 (50%)
Completed Graduate School	4 (13.3%)	1 (16.7%)
Dietary restraint: total score	28.3 (4.9)	26.7 (5.9)	0.487
Standardized score	2.83 (0.49)	2.67 (0.59)
Emotional eating: total score	31.1 (8.9)	30.2 (14.5)	0.841
Standardized score	2.39 (0.69)	2.32 (1.12)
External eating: total score	30.5 (5.2)	26.0 (5.9)	0.068
Standardized score	3.05 (0.52)	2.60 (0.59)
Weight (kg)	68.18 (10.46)	68.15 (5.67)	0.995
BMI (kg/m2)	25.29 (3.72)	24.86 (0.78)	0.798
BMI Category			0.650
20–24.9 kg/m2	18 (60%)	3 (50%)
≥ 25 kg/m2	12 (40%)	3 (50%)

Results are listed as mean (standard deviation) unless otherwise specified. Note that multiple races can be selected, hence figures may not add up to 100%. Dietary restraint and external eating total scores can range from 10 to 50, emotional eating total scores can range from 13 to 65; standardized scores for all subscales range from 1 to 5

Eating behavior and diet outcomes

Twenty-five participants had food records available to assess HEI and caloric intake. There was a statistically significant decrease in emotional eating from baseline to end of TRE, representing a medium effect (p = 0.009, Cohen’s d = − 0.470). There were no significant changes in dietary restraint, external eating, or diet quality as assessed by HEI from baseline to end of TRE (see Table 3). Total calorie intake per day decreased from 1637 ± 326 kcal/day at baseline to 1333 ± 391 kcal/day at the end of TRE, representing a medium effect (p < 0.001, Cohen’s d = − 0.750).

Table 3.

Change in eating behavior and diet outcomes

	Baseline Mean (SD)	End of TRE Mean (SD)	p	Cohen’s d
Dietary restraint	28.3 (5.0)	29.0 (6.1)	0.292	0.199
Emotional eating	31.3 (9.0)	28.9 (8.1)	0.009	− 0.470
External eating	30.6 (5.3)	30.0 (6.1)	0.200	− 0.159
Diet quality (HEI score)	55.1 (12.4)	53.8 (11.0)	0.675	− 0.092
Kilocalories per day	1637 (326)	1333 (391)	< 0.001	− 0.750

Body weight and composition

There was a statistically significant decrease in weight from baseline to end of TRE, representing a medium effect (p < 0.001, Cohen’s d = − 0.649). Mean weight loss was 0.52 ± 1.1 kg in participants with BMI < 25 kg/m² and 1.24 ± 1.2 kg in participants with BMI ≥ 25 kg/m². The mean percentage of body weight lost was 1.06 ± 1.67%. The precision (estimated by coefficient of variation across repeat measurements) for total percent body fat assessed by DXA was 0.85%, which corresponds to a least significant change of 2.36% required to detect a difference in total body fat percentage. While change in total body fat percentage was not significant (p = 0.266), there was a significant decrease in visceral fat mass (p = 0.026, Cohen’s d = − 0.452, see Table 4).

Table 4.

Change in weight and body composition outcomes

	Baseline Mean (SD)	End of TRE Mean (SD)	p	Cohen’s d	% change from baseline Mean (SD)
Weight (kg)	67.3 (10.1)	66.5 (9.8)	< 0.001	− 0.649	− 1.06 (1.67)
Total % body fat	39.3 (5.2)	39.0 (5.2)	0.266	− 0.250	− 0.41 (2.69)
Visceral fat mass (g)	312 (153)	293 (131)	0.026	− 0.452	− 3.74 (12.64)

Physical activity and sleep

Mean daily minutes of MVPA and sleep did not differ statistically between baseline and the TRE period. The weighted mean for daily minutes of MVPA was 188 ± 47 min during the baseline week and 191 ± 32 min during TRE (p = 0.757, n = 22 with data available for MVPA). Mean sleep time was 6.4 ± 1.1 h during the baseline week and 7.1 ± 2.6 h during TRE (p = 0.091).

Ecological momentary assessment

On average, participants completed 87% of possible EMA surveys during the baseline week, 86% of surveys during week 1 of TRE, and 87% of surveys during week 4 of TRE. Across all weeks, survey completion declined throughout the day: EMA survey 1: 90–93% completed, survey 2: 92–95%, survey 3: 88–91%, survey 4: 85–89%, survey 5: 66–73%. There were no missing data in the recorded surveys.

In total, 2642 surveys were completed, with eating reported on 55% of EMA surveys. Participants reported eating due to being a little bit hungry on 40% of eating occasions and very hungry on 46% of eating occasions. Eating in the absence of hunger occurred on 14% of eating occasions. Reasons for eating differed between EAH occasions and eating with hunger occasions (see Fig. 3). When participants reported EAH, they reported higher levels of stress eating (4.8%) and eating when bored, sad, or lonely (6.8%). When participants reported eating with hunger, they typically reported low amounts of emotional eating (0.5% of eating occasions were due to stress, 0.7% were eating when bored, sad, or lonely). Similarly, participants reported higher levels of impulsivity and external eating when EAH compared to when eating with hunger. For example, “it was an impulse” was chosen on 10.1% of EAH occasions vs 1.0% of eating with hunger occasions, and “it looked good/enjoyable” was chosen on 28.5% of EAH occasions vs 14.0% of eating with hunger occasions.

Fig. 3.

Participant selected reasons for eating by hunger status at eating occasion

Eating in the absence of hunger

The first GLMM (Model 1) assessed presence of EAH during eating occasions, with study week entered as a fixed factor (baseline, week 1 of TRE, and week 4 of TRE) and a random intercept and slope to account for difference in participants’ baseline levels of EAH and to allow for individual differences in change in EAH over time. There was an overall significant effect of study week (p < 0.001). As compared to baseline, participants demonstrated 52% lower odds of eating in the absence of hunger during week 1 of TRE (OR = 0.48, 95% CI 0.32–0.71, p < 0.001). Similarly, the odds of eating in the absence of hunger were 54% lower during week 4 of TRE as compared to baseline (OR = 0.46, 95% CI 0.29–0.71, p < 0.001). Predicted probabilities of EAH were 0.162 (95% CI 0.097–0.257) at baseline, 0.084 (95% CI 0.047–0.145) at week 1, and 0.081 (95% CI 0.044–0.144) at week 4 (see Fig. 4).

Fig. 4.

Predicted probability of eating in the absence of hunger during eating occasions. Chart displays mean predicted probabilities and standard errors

The second and third GLMMs assessed within-day changes in EAH. The second GLMM (Model 2), which assessed the presence of EAH during eating occasions, found the interaction of study week and EMA survey number was not significant (p = 0.385), indicating that out of occasions when participants ate, changes in EAH across study weeks were not different based on time of day. In the third GLMM (Model 3), which assessed the outcome of how frequently EAH occurred out of all EMA surveys, including times when eating did not occur, the interaction of study week by EMA survey time was significant (p = 0.04), indicating that changes in EAH between study weeks were different depending on the time of day (see Fig. 5).

Fig. 5.

a Predicted probabilities for eating in the absence of hunger over time for each EMA survey. b Within-day variation in predicted probability of eating in the absence of hunger by study week. Graph displays predicted probabilities of EAH across all surveys, including times at which eating did not occur. Error bars represent standard error

Examination of simple contrasts from Model 3 showed EAH was significantly lower during both weeks 1 and 4 of TRE compared to baseline at EMA survey 1 (week 1 p = 0.026, week 4 p = 0.019), EMA survey 2 (week 1 p = 0.033, week 4 p = 0.033), and EMA survey 5 (week 1 p = 0.002, week 4 p = 0.002). EMA survey 1 was administered at the start of the eating window and asked about eating since going to bed the night before, EMA survey 2 was administered in the first 4 h of the eating window and asked about eating that had occurred in the last 2 h and EMA survey 5 was given at bedtime and asked about eating that occurred since the close of the eating window. There were no differences in EAH between baseline and either TRE week for the EMA surveys that occurred later in the eating window (EMA surveys 3 and 4, see Table 5).

Table 5.

Simple contrasts of differences in eating in the absence of hunger across study weeks for each EMA survey

EMA survey number	Week of study	Contrast estimate	Standard error	Adjusted significance	95% Confidence interval
Survey 1	Week 1 of TRE—baseline	− 0.042	0.019	0.026	− 0.080	− 0.005
	Week 4 of TRE—baseline	− 0.050	0.019	0.019	− 0.093	− 0.007
Survey 2	Week 1 of TRE—baseline	− 0.069	0.029	0.033	− 0.133	− 0.004
	Week 4 of TRE—baseline	− 0.070	0.029	0.033	− 0.136	− 0.005
Survey 3	Week 1 of TRE—baseline	− 0.027	0.028	0.579	− 0.090	0.037
	Week 4 of TRE—baseline	− 0.028	0.030	0.579	− 0.094	0.039
Survey 4	Week 1 of TRE—baseline	− 0.018	0.028	0.763	− 0.081	0.044
	Week 4 of TRE—baseline	− 0.017	0.030	0.763	− 0.081	0.048
Survey 5	Week 1 of TRE—baseline	− 0.153	0.047	0.002	− 0.260	− 0.047
	Week 4 of TRE—baseline	− 0.145	0.047	0.002	− 0.241	− 0.050

Sequential adjusted Sidak significance level: 0.05

Discussion

In this single-arm trial, emerging adult women with moderate–high levels of dietary restraint who followed a TRE intervention experienced decreases in emotional eating and eating in the absence of hunger (EAH) and no changes in external eating or dietary restraint. In addition, visceral fat, a known risk factor for CVD [46–48], decreased between baseline and the end of the TRE intervention, suggesting that TRE could help mitigate cardiometabolic risk factors. While caloric intake and body weight decreased, diet quality remained unchanged. When considering all times surveyed, including when eating did not occur, EAH was lower during the fasting window and during the first portion of the eating window during TRE compared to baseline, while there were no changes later in the eating window.

Both emotional eating and EAH decreased during TRE. Participants reported high adherence to TRE and thus greatly reduced their eating during the evening, the time of day at which baseline ecological momentary assessment (EMA) data showed EAH was most likely to occur. EMA also indicated a relationship between emotional eating and EAH: eating events that occurred in the absence of hunger had higher levels of emotional eating compared to eating events that occurred with hunger. Contrary to our hypothesis, EAH did not increase during the eating window, indicating that eating for non-physiological reasons did not simply shift to earlier in the day. These quantitatively measured reductions in emotional eating and EAH are consistent with qualitative evaluations of TRE studies showing reduced nighttime snacking and mindless eating [49, 50].

We did not observe changes in dietary restraint, a finding consistent with other TRE studies in adults with high body weight [51, 52] and emerging adult women with polycystic ovary syndrome [53]. In contrast to the present study, none of these studies had inclusion criteria around baseline restraint scores. One used the DEBQ and reported a mean baseline DEBQ restraint score of 14 ± 6 [51]. The present study required participants to endorse at least a moderate level of dietary restraint, and baseline restraint scores were 28.3 ± 5. Taken together, these preliminary results suggest that participants did not experience adverse changes to eating behavior or cognitions following a TRE intervention.

Despite being told they were not required to change the types or amount of food consumed, participants reported an average reduction of 304 cal per day (18.6% of baseline intake). This level of caloric reduction is consistent with other TRE studies, which report spontaneous reductions in energy intake of up to 20% [54, 55]. Although the overall quantity of caloric intake declined, diet quality did not change from baseline to end of TRE, consistent with other TRE intervention studies which have found no changes in diet quality, eating frequency, or macronutrient distribution following a TRE intervention [49, 51]. Exploration of how the different components of HEI change with TRE is an area for future research, and may inform strategies for guiding participants to improve their diet quality while engaging in a TRE intervention. The impact of interventions that combine TRE with other healthful dietary patterns such as the Mediterranean diet is also an area that warrants further exploration [56].

Participants had significant decreases in weight and visceral fat, but no changes in overall body fat. The modest mean level of weight loss observed in this group (mean change of 0.8 kg or 1.75 lbs. over 4 weeks, representing a mean of 1.06% of body weight) is consistent with the decrease in caloric intake. TRE interventions commonly produce a weight loss of approximately 3–4% of body weight; however, many TRE studies are longer than 4 weeks or occur in higher weight individuals [54, 55]. The level of weight loss observed in the present study is consistent with a 4-week TRE study in emerging adults with a mean BMI in the “healthy” range, which found a significant 1.4 kg decrease in weight for the group that ended their eating window by 8 pm [36]. While weight loss is not necessarily beneficial for all participants, we observed greater weight loss in participants with higher body weight. Weight loss was also modest and associated with a reduction in visceral fat. Visceral fat losses have been observed in other studies of TRE interventions conducted in populations with metabolic dysfunction (e.g., metabolic syndrome) [53, 57, 58]. It is possible that the intervention was not long enough to produce total body fat loss, or that the 10-h eating window was insufficient to prompt fat loss. A study assessing different eating window lengths found fat loss only in the group with an 8-h eating window, while 10- and 12-h eating window groups did not experience fat loss [59].

Finally, while overall sleep and physical activity levels did not change statistically, there was a clinically meaningful change in sleep duration, such that the mean sleep duration of the group increased during TRE to a level that meets the sleep recommendation of 7–9 h per night [60]. The time at which activity and sleep behaviors occurred in the 24-h cycle may also have been impacted. This is a promising area for future research.

Strengths and limitations

Strengths of this study include use of EMA to capture within-day measures of eating behavior throughout the study, DXA for measurement of body composition, and remote time-stamped photo-assisted food records. There are some limitations that should be noted. As this was a single-arm trial without a control group, we cannot be certain that changes seen in the study were caused by the TRE intervention and not by other confounding factors. Participants were enrolled across 18 months at a relatively consistent rate, decreasing the probability that seasonal variations or specific events (i.e., final exams) may have impacted outcomes. Our sample was predominantly college students identifying as White and Asian, and thus results may not be generalizable to other demographic groups. In addition, selection bias could have occurred, with individuals who were interested in TRE being more likely to join and complete the study. However, as individuals who are interested in TRE are likely to try it in community settings, results may be applicable to clinical practice. Participants withdrawing from the study reduced statistical power and may have limited our ability to detect smaller changes in outcomes. Finally, we did not standardize the study intervention based on participants’ menstrual cycle phase, and hormonal fluctuations could have impacted dietary intake apart from the intervention.

Randomized controlled trials are needed to confirm these findings. While our preliminary results suggest decreases in factors associated with CVD risk, such as emotional eating, eating in the absence of hunger, and visceral fat, CVD risk is determined by multiple behavioral and physiological factors [61]. Studies that assess additional physiological outcomes such as blood lipids, glucose, and blood pressure along with behavioral outcomes would provide a more complete assessment of the relationship between TRE interventions and CVD risk. Caution is needed when interpreting the finding that TRE did not increase dysregulated eating. Given the lack of a control group and the short duration of the intervention, any definitive conclusions about the absence of negative psychological effects are premature. Studies that examine moderating variables could provide additional insights on differential impacts of TRE on diverse populations. While TRE is not a ‘diet’ and restrictions are not placed on the type or amount of food consumed, restrictions are placed on the time a person is ‘allowed’ to eat and restrictive interventions are contraindicated for individuals at risk for eating disorders [62].

In conclusion, data from this study suggest that 4-week TRE intervention did not have deleterious effects on eating behaviors in emerging adult women. Participants experienced decreased emotional eating and eating in the absence of hunger. Modest calorie reductions and loss of weight and visceral fat were also observed. These results provide preliminary support for the use of TRE as a strategy for decreasing CVD risk in emerging adult women.

What is already known on this subject?

TRE confers cardiometabolic benefits. However, the impact of TRE on dysregulated eating is less well-understood. It is essential to understand the impact of TRE on dysregulated eating, as increases in dysregulated eating increase cardiometabolic risk.

What this study adds?

This study shows that following a TRE intervention decreased emotional eating, eating in in the absence of hunger and visceral fat mass among emerging adult women with moderate–high dietary restraint.

Author contributions

Conceptualization and Methodology: Diane Vizthum, Carrie P. Earthman, Freda Patterson, Melissa M. Melough, Carly R. Pacanowski; Investigation: Diane Vizthum, Carrie P. Earthman; Formal analysis: Diane Vizthum, Carrie P. Earthman, Freda Patterson, Melissa M. Melough, Carly R. Pacanowski; Writing—original draft preparation: Diane Vizthum; Writing—review and editing: Carrie P. Earthman, Freda Patterson, Melissa M. Melough, Kelly C. Allison, Carly R. Pacanowski; Funding acquisition: Diane Vizthum, Carrie P. Earthman, Freda Patterson, Melissa M. Melough, Carly R. Pacanowski; Resources: Carrie P. Earthman, Freda Patterson; Supervision: Kelly C. Allison, Carly R. Pacanowski.

Funding

This work was funded by the Academy of Nutrition and Dietetics Foundation’s 2024 Commission on Dietetic Registration Emerging Researcher Grant.

Data availability

Data are available upon request to the corresponding author.

Declarations

Ethics approval

This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the University of Delaware’s Institutional Review Board (Sept 5, 2023/2089190).

Consent to participate

Informed consent was obtained from all individual participants included in the study.

Competing interests

D.V., C.P.E., F.P., M.M.M., and C.R.P. have no relevant financial or non-financial interests to disclose. K.C.A. received grant funding from an Investigator Initiated Study from Novo Nordisk that was unrelated to the current study.