Effectiveness of Early Time-Restricted Eating for Weight Loss, Fat Loss, and Cardiometabolic Health in Adults With Obesity: A Randomized Clinical Trial

Key Points

Question  Is early time-restricted eating more effective than eating over a period of 12 or more hours for losing weight and body fat?

Findings  In a randomized clinical weight-loss trial involving 90 adults with obesity, early time-restricted eating was more effective for losing weight (−6.3 kg) than eating over a window of 12 or more hours (−4.0 kg) but not for losing body fat (−4.7 vs −3.4 kg). In a secondary analysis of completers, early time-restricted eating was more effective for losing weight and body fat.

Meaning  Early time-restricted eating was more effective for weight loss than eating over a window of 12 or more hours; larger studies are needed on fat loss.

Importance  It is unclear how effective intermittent fasting is for losing weight and body fat, and the effects may depend on the timing of the eating window. This randomized trial compared time-restricted eating (TRE) with eating over a period of 12 or more hours while matching weight-loss counseling across groups.

Objective  To determine whether practicing TRE by eating early in the day (eTRE) is more effective for weight loss, fat loss, and cardiometabolic health than eating over a period of 12 or more hours.

Design, Setting, and Participants  The study was a 14-week, parallel-arm, randomized clinical trial conducted between August 2018 and April 2020. Participants were adults aged 25 to 75 years with obesity and who received weight-loss treatment through the Weight Loss Medicine Clinic at the University of Alabama at Birmingham Hospital.

Interventions  All participants received weight-loss treatment (energy restriction [ER]) and were randomized to eTRE plus ER (8-hour eating window from 7:00 to 15:00) or control eating (CON) plus ER (≥12-hour window).

Main Outcomes and Measures  The co–primary outcomes were weight loss and fat loss. Secondary outcomes included blood pressure, heart rate, glucose levels, insulin levels, and plasma lipid levels.

Results  Ninety participants were enrolled (mean [SD] body mass index, 39.6 [6.7]; age, 43 [11] years; 72 [80%] female). The eTRE+ER group adhered 6.0 (0.8) days per week. The eTRE+ER intervention was more effective for losing weight (−2.3 kg; 95% CI, −3.7 to −0.9 kg; P = .002) but did not affect body fat (−1.4 kg; 95% CI, −2.9 to 0.2 kg; P = .09) or the ratio of fat loss to weight loss (−4.2%; 95% CI, −14.9 to 6.5%; P = .43). The effects of eTRE+ER were equivalent to reducing calorie intake by an additional 214 kcal/d. The eTRE+ER intervention also improved diastolic blood pressure (−4 mm Hg; 95% CI, −8 to 0 mm Hg; P = .04) and mood disturbances, including fatigue-inertia, vigor-activity, and depression-dejection. All other cardiometabolic risk factors, food intake, physical activity, and sleep outcomes were similar between groups. In a secondary analysis of 59 completers, eTRE+ER was also more effective for losing body fat and trunk fat than CON+ER.

Conclusions and Relevance  In this randomized clinical trial, eTRE was more effective for losing weight and improving diastolic blood pressure and mood than eating over a window of 12 or more hours at 14 weeks.

Trial Registration  ClinicalTrials.gov Identifier: NCT03459703

Intermittent fasting (IF) is the practice of alternating eating and extended fasting. In recent years, IF has been touted for losing weight and body fat. Indeed, IF can decrease body fat and preserve lean mass in animals and humans.^1-11 Moreover, a few clinical trials have reported that IF is better for losing weight and/or body fat than continuous energy restriction in the short term.^3,6,12-16 However, the literature is mixed, and there is no definitive evidence that IF selectively burns body fat while sparing muscle tissue.

One form of intermittent fasting that is particularly promising is time-restricted eating (TRE), which we define as eating within a consistent window of 10 hours or less and fasting for the rest of the day.¹⁷ Studies have shown that TRE prevents and reverses diet-induced obesity in rodents.^18-20 Adults who adhere to TRE typically lose 1% to 4% of their body weight within several weeks.^21-32 Further, TRE increases fat oxidation³³ and can improve cardiometabolic end points, such as insulin sensitivity and blood pressure, even when calorie intake is matched to the control group.^17,33-36

Although promising, most trials on TRE are small or single arm or used a weak control group. Therefore, we conducted a weight-loss randomized clinical trial comparing TRE with eating over a window of 12 or more hours, where both groups received identical weight-loss counseling. We tested a version of TRE called early TRE (eTRE), which involves stopping eating in the afternoon and fasting for the rest of the day. Because key circadian rhythms in metabolism—such as insulin sensitivity and the thermic effect of food—peak in the morning, eTRE may confer additional benefits relative to other forms of TRE.³⁷ We hypothesized a priori that eTRE would be more effective for losing weight and body fat and improving cardiometabolic health than eating over a window of 12 or more hours and that participants would adhere to eTRE about 5 days per week.

New patients with obesity at the Weight Loss Medicine Clinic of the University of Alabama at Birmingham (UAB) Hospital were recruited between August 2018 and December 2019 by direct email, clinic newsletter, and physician referral. Applicants were eligible if they were aged 25 to 75 years, had a body mass index (BMI; calculated as weight in kilograms divided by height in meters squared) between 30.0 and 60.0, and did not have diabetes or a severe or unstable chronic medical condition. Additional eligibility criteria are listed in the eMethods in Supplement 1. Participants self-reported their race, ethnicity, and sex. All participants provided written informed consent, and the study was approved by UAB’s Institutional Review Board (IRB-300001207). The trial protocol and statistical analysis plan appear in Supplement 2 and Supplement 3, respectively.

The study was a 14-week parallel-arm, randomized controlled weight-loss trial. Participants were randomized to follow eTRE (8-hour eating window between 7:00 and 15:00) or a control (CON) eating schedule (a self-selected ≥12-hour window), which was designed to mimic US median meal timing habits.³⁸ Participants were instructed to follow their assigned eating schedule at least 6 days per week. Aside from when participants ate, all other intervention components were matched across groups. Randomization was performed by the statistician in a 1:1 allocation ratio, with stratification by sex, race (Black vs not Black), and baseline physical activity level (≤2 days/week vs ≥3 days/week of exercise of any duration or intensity), using block sizes of 2 in the software program R.

All participants received weight-loss counseling involving energy restriction (ER) at the UAB Weight Loss Medicine Clinic. In brief, participants received one-on-one counseling from a registered dietitian at baseline (60-minute session) and at weeks 2, 6, and 10 (30-minute sessions). Participants were counseled to follow a hypocaloric diet (500 kcal/d below their resting energy expenditure) and exercise 75 to 150 minutes per week, depending on their baseline physical activity. Participants were also instructed to attend at least 10 group classes. See eMethods in Supplement 1 for more details.

The co–primary outcomes were weight loss and fat loss. The secondary outcomes were fasting cardiometabolic risk factors. Additional outcomes included adherence, satisfaction with the eating windows, food intake, physical activity, mood, and sleep. All week 0 and 14 outcomes except adherence and food intake were measured in the morning following a water-only fast of at least 12 hours. In addition, we measured body weight in the nonfasting state in the clinic every 2 weeks throughout the trial.

Body composition was measured using dual x-ray absorptiometry (DEXA [iDXA; GE-Lunar Radiation Corporation]) and analyzed using enCORE software, version 15 (GE Healthcare). Fat loss was assessed in 2 ways: as the ratio of fat loss to weight loss (primary fat loss end point) and as the absolute change in fat mass (secondary fat loss end point). The former was used to test the hypothesis that IF selectively burns more body fat and less lean tissue per kilogram of weight lost (as a proxy for “selective fat loss”), while the latter was used to represent total fat loss. To accurately assess the former end point, we limited the analysis to completers who lost at least 3.6 kg (see eMethods in Supplement 1).

Fasting blood pressure, glucose levels, insulin levels, homeostatic model assessment for insulin resistance (HOMA-IR), HOMA for β-cell function (HOMA-β), hemoglobin A_1c level, and plasma lipid levels were measured using standard procedures (see eMethods in Supplement 1).

Participants reported when they started and stopped eating daily through surveys administered via REDCap (Research Electronic Data Capture) software.^39,40 Participants were classified as adherent if they followed their assigned eating window within 30 minutes. Days with missing surveys were considered nonadherent.

Energy intake and macronutrient composition were measured by 3-day food record using the Remote Food Photography Method.⁴¹ We also conducted a post hoc analysis of energy intake using weight-loss modeling techniques (see eMethods in Supplement 1). We measured physical activity, mood, sleep, and satisfaction with the eating window using the Baecke Physical Activity Questionnaire, the Profile of Moods–Short Form (POMS-SF), the Patient Health Questionnaire-9 (PHQ-9), the Munich Chronotype Questionnaire (MCTQ), the Pittsburgh Sleep Quality Index (PSQI), and a 5-point Likert scale, respectively (see eMethods in Supplement 1).

The trial was statistically powered to detect a 10.0 plus or minus 13.4% difference in the ratio of fat loss to weight loss. We decided to assess the ratio of fat loss to weight loss only in completers who lost at least 3.6 kg because otherwise technical errors associated with DEXA can produce spurious values greater than 100% to 200% or less than 20% when participants lose marginal amounts of fat and/or fat-free mass (see eMethods in Supplement 1). Assuming 30% of participants would drop out or not lose at least 3.6 kg (8 lb), an estimated 86 enrollees were needed to have 60 completers who lost sufficient weight.

Analyses were performed in R, version 4.0.3 (R Foundation for Statistical Computing) using 2-sided tests with α = .05. All analyses were intention-to-treat, except that the ratio of fat loss to weight loss and questionnaire data were analyzed in completers only. End points with 3 or more repeated measures included body weight and adherence and were analyzed using linear mixed models. All other end points were analyzed using multiple imputation by chained equations, followed by linear regression. Between-group analyses were adjusted for age, race (Black vs non-Black), and sex (male vs female), while baseline data and within-group changes were analyzed using independent t tests. Following our preregistered statistical plan, we also performed a secondary analysis in completers using the same statistical methods. See eMethods in Supplement 1 for more statistical details.

We screened 656 people and enrolled 90 participants (Figure 1). Participants had a mean (SD) BMI of 39.6 (6.7) and mean age of 43 (11) years. Seventy-two participants (80%) were female; 2 (2%) were Asian, 30 (33%) were Black, 56 (62%) were White, and 2 (2%) reported being of more than 1 race; and 2 (2%) were of Hispanic ethnicity, 85 (94%) were not of Hispanic ethnicity, and 3 (3%) had unknown or not reported ethnicity (Table 1). The attrition rate was similar between groups: 9 (20%) and 11 (24%) participants dropped out of the CON+ER and eTRE+ER groups, respectively (P = .80). Two participants in the eTRE+ER group withdrew owing to difficulty following the eating schedule, whereas none in the CON+ER group withdrew. Adverse events in both groups were mild (see eAppendix in Supplement 1). Unfortunately, because of the COVID-19 pandemic, we were unable to collect postintervention data on primary and secondary outcomes in 11 participants (see eMethods in Supplement 1). As a result, 59 participants (66%) completed all aspects of the study (see eTable 1 in Supplement 1 for a comparison of completers vs noncompleters).

During the intervention, the eTRE+ER group ate within a mean (SD) time period of 7.6 (0.8) hours, while the CON+ER group ate over a mean (SD) time period of 12.3 (0.8) hours—a difference of about 4.8 hours (P < .001). Both groups ate breakfast at a similar time, but the eTRE+ER group finished eating at mean (SD) 15:35 (0:34) vs 20:25 (1:02) in the CON+ER group (P < .001; Figure 2A). The eTRE+ER group adhered a mean (SD) of 6.0 (0.8) days per week, which was lower than the CON+ER group (6.3 [0.8] days/week; P = .03), and adherence in the eTRE+ER group declined by 0.4 days per week over the 14-week intervention (P = .001; Figure 2B). Satisfaction with the eating window was similar between groups (−0.3; 95% CI, −0.8 to 0.2; P = .18; Figure 2C). After completing the intervention, 12 (41%) of the eTRE+ER group wanted to continue eTRE (vs 2 [7%] in the CON+ER group). Eight (28%) planned to eat within the originally prescribed 7:00 to 15:00 window, while the remainder wanted to use a different eTRE window (Figure 2D; see eMethods in Supplement 1).

Both the eTRE+ER group (−6.3 kg [−5.7%]; 95% CI, −7.4 to −5.2 kg; P < .001) and CON+ER group (−4.0 kg [−4.2%]; 95% CI, −5.1 to −2.9 kg; P < .001) achieved clinically meaningful weight loss (Figure 3). The eTRE+ER group lost an additional 2.3 kg of body weight (95% CI, −3.7 to −0.9 kg; P = .002) relative to the CON+ER group. However, there were no statistically significant differences in absolute fat loss (−1.4 kg; 95% CI, −2.9 to 0.2 kg; P = .09) or the ratio of fat loss to weight loss (n = 41; −4.2%; 95% CI, −14.9 to 6.5%; P = .43). There were also no statistically significant differences in the changes in fat-free mass, trunk fat, visceral fat, waist circumference, or appendicular lean mass (Table 2).

The eTRE+ER intervention lowered diastolic blood pressure by an additional 4 mm Hg (95% CI, −8 to 0 mm Hg; P = .04) relative to CON+ER (Table 2). There were no statistically significant differences in systolic blood pressure, heart rate, glucose levels, insulin levels, HOMA-IR, HOMA-β, hemoglobin A_1c level, or plasma lipid levels (Table 2).

There were no between-group differences in self-reported physical activity, energy intake (1 kcal/d; 95% CI, −251 to 253 kcal/d; P = .99), or macronutrient composition (eTable 2 in Supplement 1). However, weight-loss modeling⁴² of all participants with at least 2 weight measurements (n = 77) indicated that eTRE reduced energy intake by an additional 214 kcal/d (95% CI, −416 to −12 kcal/d; P = .04) relative to the control eating window.

The eTRE+ER intervention was more effective at improving total mood disturbances, as well as mood subscores for vigor-activity, fatigue-inertia, and depression-dejection (eFigure 1 in Supplement 1). All other mood and sleep end points were similar between groups (eFigures 1 and 2 in Supplement 1).

In a preregistered secondary analysis of completers (eTable 3 in Supplement 1), eTRE was more effective for losing weight (−2.3 kg; 95% CI, −3.9 to −0.7 kg; P = .006), body fat (−1.8 kg; 95% CI, −3.6 to 0.0 kg; P = .047), and trunk fat (−1.2 kg; 95% CI, −2.2 to −0.1 kg; P = .03). Among secondary outcomes, eTRE+ER was more effective at lowering diastolic blood pressure (−5 mm Hg; 95% CI, −9 to −1 mm Hg; P = .01). All other primary and secondary outcomes were similar between groups (eTable 3 in Supplement 1).

We conducted a randomized weight-loss trial comparing TRE with eating over a period of 12 or more hours where both groups received the same weight-loss counseling. Our data suggest that eTRE is feasible, as participants adhered 6.0 days per week on average, and most participants adhered at least 5 days per week. Despite the challenges of navigating evening social activities and occupational schedules, adherence to eTRE was similar to that of other TRE interventions (approximately 5.0-6.4 days/week),^{21-23,25,26,43-45} and satisfaction was similar between groups. Importantly, participants in the eTRE+ER group experienced greater improvements in mood—including fatigue, vigor, and feelings of depression/dejection—which may have helped them adhere to eTRE. Furthermore, we found that eTRE was acceptable for many patients. About 41% of completers in the eTRE+ER group planned to continue practicing eTRE after the study concluded.

The key finding of this study is that eTRE was more effective for losing weight than eating over a period of 12 or more hours. In our trial, the eTRE group lost an additional 2.3 kg relative to the control group, an approximately 50% improvement in weight loss. For comparison, prior studies are about evenly divided on whether TRE reduces body weight^{15,16,21,23,25,43,44,46-53} and are mixed for body fat,^{9,11,16,23,34,43,46,49-51,53-56} while studies that shift food intake to the morning and/or earlier in the daytime have more consistently reported weight loss.^57-62 Of note, a somewhat larger study recently published in the New England Journal of Medicine by Liu et al⁵³ reported that eTRE was not better for losing weight. However, our study had better post hoc statistical power owing to less variability in weight loss. Our 95% CI was narrower than and wholly contained within the other trial’s 95% CI. Therefore, our results are not incompatible. Furthermore, our eTRE group extended their daily fasting by twice as much, fasting an extra 4.8 hours per day vs only a modest 2.3-hour change in the Liu et al⁵³ study. The magnitude of the weight-loss effect we observed was equivalent to reducing energy intake by an additional 214 kcal/d and is in line with 2 recent meta-analyses reporting that TRE has modest to moderate effects on body weight.^31,32 Though we could not trace this energy deficit to changes in physical activity or food intake, we suspect that eTRE did reduce energy intake, but we were unable to detect it owing to the well-known limitations of accurately assessing food intake via self-report. Most previous studies report that TRE reduces energy intake and does not affect physical activity.^{21,24,25,27,51,52,55}

On the other hand, we found no evidence of selective fat loss, as measured by the ratio of fat loss to weight loss. Also, total fat loss was not statistically significant in the main intention-to-treat analysis. Our finding of a difference in weight loss but not fat loss was likely due to lower statistical power because DEXA scans were performed only twice (whereas body weight was measured 8 times) and using a conservative imputation approach. In a secondary analysis of completers, eTRE was indeed better for losing body fat and trunk fat than eating over a window of 12 or more hours. The eTRE intervention increased fat loss by an additional 1.8 kg or 18 g/d. Our finding is consistent with a study reporting that participants burned an extra 15 g/d of fat when they ate 4.5 hours earlier in the day.⁶³ Importantly, we found no evidence that extending the daily fasting period negatively affected lean mass, which is consistent with most studies on TRE.^{9,11,34,46,54,55} Although a large randomized clinical trial published in JAMA Internal Medicine by Lowe et al⁴³ found that practicing TRE by skipping breakfast decreased appendicular lean mass, we had a larger sample size of DEXA scans and did not observe any deleterious effects.

The eTRE intervention was also more effective than eating over a period of 12 or more hours for lowering diastolic blood pressure. The effects were clinically significant and on par with those of the DASH (Dietary Approaches to Stop Hypertension) diet⁶⁴ and endurance exercise.⁶⁵ Although the effect was identical for systolic blood pressure, it was not statistically significant owing to the larger variance, suggesting that larger studies are needed on systolic blood pressure. For comparison, 1 previous controlled feeding study reported that eTRE reduces blood pressure,¹⁷ while other TRE studies are mixed but lean null.^{11,16,17,21-23,25,27,28,34,43,50,66,67} Studies reporting improvements tend to be longer and/or involve eating earlier in the day relative to the control group. Indeed, blood pressure has a pronounced circadian rhythm,⁶⁸ and circadian misalignment elevates blood pressure in humans.⁶⁹ Though other factors—such as fasting natriuresis or changes in sympathetic tone—could also explain these effects.

The eTRE intervention was not more effective for improving other fasting cardiometabolic end points. Our results are at odds with most studies on eTRE^17,35,36,70 or eating earlier in the daytime,^{57,58,61,62,71-73} which usually report improvements in postprandial or 24-hour glucose levels, insulin sensitivity, and/or fasting insulin levels. However, studies on other versions of TRE report more mixed results.^{23,34,35,51,52,74,75} We acknowledge the limitation that we did not measure glycemic end points in the postprandial state, which are more responsive to dietary interventions. We also had larger variability in fasting insulin level relative to our previous trial.¹⁷ For plasma lipid levels, our results are consistent with most studies on TRE, which report no effects.^{9,11,17,21,23,25-28,34,35,43,47,48,52,54,66,70,74}

Our study has a few limitations, including being modest in duration, enrolling mostly women, and not achieving our intended sample size, partly owing to the COVID-19 pandemic. Also, we measured physical activity by self-report, not by accelerometry, which may have limited our ability to detect differences in physical activity between groups. Finally, we measured cardiometabolic end points only in the fasting state. Future research should investigate glycemic end points in the postprandial state or over a 24-hour period.

In this randomized clinical trial, eTRE was more effective for losing weight and lowering diastolic blood pressure than eating over a period of 12 or more hours at 14 weeks. The eTRE intervention may therefore be an effective treatment for both obesity and hypertension. It also improves mood by decreasing fatigue and feelings of depression-dejection and increasing vigor, and those who can stick with eTRE lose more body fat and trunk fat. However, eTRE did not affect most fasting cardiometabolic risk factors in the main intention-to-treat analysis.

This trial also lays important groundwork for future IF research. In this trial, eTRE produced clinically meaningful improvements in body fat, trunk fat, and systolic blood pressure that fell within the statistical trend range, with modest to moderate effect sizes (Cohen d = 0.32-0.39). Therefore, future clinical trials will need to enroll much larger sample sizes—up to approximately 300 participants—to determine whether IF affects body composition and cardiometabolic health. Future studies should investigate whether the timing and duration of the eating window affect these results, as well as determine who can adhere to eTRE vs who cannot and would instead benefit from other meal-timing interventions. The eTRE intervention should be further tested as a low-cost, easy-to-implement approach to improve health and treat disease.

Accepted for Publication: May 13, 2022.

Published Online: August 8, 2022. doi:10.1001/jamainternmed.2022.3050

Corresponding Author: Courtney M. Peterson, PhD, University of Alabama at Birmingham, 1675 University Blvd, Webb 644, Birmingham, AL 35233 (cpeterso@uab.edu).

Author Contributions: Drs Peterson and Richman had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Jamshed and Steger contributed equally to this work as co–first authors.

Concept and design: Salvy, Peterson.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Steger, Salvy, Peterson.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Richman, Peterson.

Obtained funding: Peterson.

Administrative, technical, or material support: Jamshed, Steger, Bryan, Hanick, Martin, Peterson.

Supervision: Warriner, Hanick, Peterson.

Conflict of Interest Disclosures: Dr Martin reported grants from the National Institutes of Health (NIH) during the conduct of the study and personal fees (scientific advisory board member) from Wondr Health outside the submitted work. Pennington Biomedical Research Center/Louisiana State University has interest in the intellectual property surrounding the Remote Food Photography Method and SmartIntake app, which were used to measure food intake, and Dr Martin is an inventor of the technology. Dr Peterson reported grants from the NIH during the conduct of the study. No other disclosures were reported.

Funding/Support: This study was supported by grants UL1 TR001419 from the National Center for Advancing Translational Sciences of the NIH and P30 DK056336 from the National Institute of Diabetes and Digestive and Kidney Diseases. Resources and support were also provided by 2 Nutrition Obesity Research Center (NORC) grants (P30 DK056336; P30 DK072476), a Diabetes Research Center (DRC) grant (P30 DK079626), an NIH Predoctoral T32 Obesity Fellowship to Mr Hanick (T32 HL105349), and the Louisiana Clinical and Translational Science Center (LA CaTS; U54 GM104940).

Role of the Funder/Sponsor: The funding provided by UL1 TR001419 was administered through the UAB Center for Clinical and Translational Science, which required a preassigned statistician to draft the statistical analysis plan as a stipulation of applying for funding. The statistician was later changed prior to beginning data analysis. The sponsors had no other roles in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Meeting Presentation: Results from preliminary analyses, which did not use linear mixed modeling, were presented at ObesityWeek 2020 and a handful of invited seminars. Full analyses, which included linear mixed models for adherence and weight loss, were conducted later.

Data Sharing Statement: See Supplement 4.

Additional Contributions: We thank the UAB Weight Loss Medicine clinic staff, and Karin Crowell, RD (Department of Medicine, UAB), especially, for their support and dedication in conducting this study. We also thank Karissa Neubig, RD (Pennington Biomedical Research Center), and Tulsi Patel, BS (UAB), for their help in measuring dietary intake and tracking adherence. Ms Crowell and Ms Neubig received no compensation beyond that of their regular employment. Ms Patel received a small stipend.