Time-restricted eating improves health because of energy deficit and circadian rhythm: A systematic review and meta-analysis

Summary Time-restricted eating (TRE) is an effective way to lose weight and improve metabolic health in animals. Yet whether and how these benefits apply to humans is unclear. This systematic review and meta-analysis examined the effect of TRE in people with overweight and obesity statuses. The results showed that TRE led to modest weight loss, lower waist circumference and energy deficits. TRE also improved body mass index, fat mass, lean body mass, systolic blood pressure, fasting glucose levels, fasting insulin levels, and HbA1c%. Subgroup analysis demonstrated more health improvements in the TRE group than the control group under the ad libitum intake condition than in the energy-prescribed condition. Eating time-of-day advantages were only seen when there was considerable energy reduction in the TRE group than the control group (ad libitum condition), implying that the benefits of TRE were primarily due to energy deficit, followed by alignment with eating time of day.


INTRODUCTION
The past 20 years have shown dramatic increases in the prevalence of overweight and obesity, as well as related chronic diseases, such as cardiovascular disease, diabetes, and some cancers related to the conditions. 1 Therefore, obesity has become a serious public concern that threatens the global population's health.Daily caloric restriction (CR), with and without exercise, is the most common treatment used to fight obesity. 2 Despite its short-term utility for weight loss, long-term CR is challenged by homeostatic physiological adaptations to weight loss, 3 acute episodes of hunger, 4 and relatively poor adherence. 57][8] TRE is considered an intermittent CR strategy compared to continuous CR since it is assumed to unintentionally reduce caloric intake by limiting the eating window during the day.As a subset of the CR strategy, TRE is superior to continuous CR and other types of intermittent fasting due to the alignment of the eating window with the time of day. 9The concomitant reduction in energy intake 10 and considerable adherence 11 may facilitate long-term weight loss and metabolic improvements. 12,13The earliest studies on time-restricted feeding (TRF) demonstrated benefits to metabolic health and lifespan in animals, 7,[14][15][16] but the extent to which these findings from animal models are translatable to humans has not been determined. 5 growing body of literature has examined the potential benefits of TRE on weight loss and metabolic health in humans, but the findings have been mixed.Although most studies confirmed that TRE could reduce body weight, the weight loss extent remains unclear.For example, 60 participants with overweight or obesity status underwent 10-h TRE for eight weeks and lost 10.7 kg, approximately 8.5% of their initial body weight. 17,180][21][22] More controversy exists on the improvements in metabolism, which has stimulated a lot of studies and led to mixed results.Some studies explored whether TRE could enhance metabolic health, 23,24 and others attempted to understand whether these benefits could exist in the absence of energy restriction or were modulated by the eating time of day. 25,26ne important consideration that may help explain the previously mixed findings is that TRE includes several intervention subtypes, which are not equally effective. 27,28Previous studies classified TRE as under isocaloric conditions or ad libitum intake conditions according to whether the energy intake was restricted and as early time restricted eating (e-TRE) or delay-time restricted eating (d-TRE) depending on whether the daily eating window was within the early or late time of day. 10,29E-TRE is generally believed to improve metabolic homeostasis by sustaining daily rhythms in the feeding and fasting cycle, while a misalignment between eating and rhythm (d-TRE) may increase hunger and impair metabolic health. 7,30,31The variation in intervention methods regarding energy intake and eating rhythm contributed to the high heterogeneity of TRE studies.
Given the inconsistent clinical findings, meta-analysis can be used to explore the actual effect of TRE on weight loss and metabolic health as it provides a more precise estimate of the treatment effect and may explain heterogeneity between studies. 32,335][36] However, the previous analyses had limitations in selecting TRE studies with poor quality or inconsistent inclusion criteria.For instance, two of them included both people with overweight status and physical activity, 35,36 and two of them contained TRE strategies with quite short intervention durations of seven or fewer days. 29,35][36] Furthermore, the number of TRE on clinical trials has greatly increased in the past two years alone, 37 providing an opportunity for examining whether, and if so, how TRE could benefit weight loss and metabolic health in humans through a systematic review and meta-analysis.
Most clinical studies of TRE have focused on its beneficial outcomes, but there has been no general agreement on the underlying mechanism of the effects of TRE on weight loss and metabolic health.Previous studies demonstrated that participants on TRE with ad libitum intake commonly reduced their calories by 7%-22%. 37Thus, it is unclear whether energy restriction and time restriction collectively contribute to weight loss and metabolic health. 38Swiatkiewicz et al. 39 proposed that reduced energy intake may account for some beneficial effects of TRE on body weight and metabolic outcomes, but one study reported improvements in insulin sensitivity and blood pressure without energy restriction or weight loss. 40Taken together, it is unclear whether energy restriction induced by TRE protocols or the alignment of the eating window with the time of day leads to healthy outcomes. 4,27,39he present study attempted to explore the mechanism of TRE for the first time through subgroup analysis in the meta-analysis by classifying TRE into isocaloric vs. ad libitum intake and e-TRE vs. d-TRE based on different intervention approaches.We assume that in the TRE strategies, energy restriction contributes more than the eating time of day to weight loss and metabolic health, supposing participants obtained more health improvements under the ad libitum intake condition with reduced energy intake compared to the isocaloric condition and e-TRE improves health status more than d-TRE only in the former condition, while the advantage of the eating time of day did not exist once energy intake was prescribed (isocaloric condition).Considering the limitations of previous meta-analyses, we conducted a systematic review and meta-analysis to examine the effect of TRE on weight loss and metabolic health in people with overweight and obesity statuses based on RCT trials and further investigated the factors that accounted for the TRE benefits.

RESULTS
Information related to the search strategy and number of participants is shown in Figure 1 and Table 1.The study characteristics are summarized in Table 2.The mean age of the participants ranged from 22.7 to 65 years, and the mean BMI ranged from 27.8 to 38.9 kg/m 2 .The eating window ranged from four to less than 12 h, and the experiment duration ranged from five to 48 weeks.Three of the studies 21,23,41 were threearmed experiments containing two intervention groups and one control group, which we separately treated as two experiments in the data extraction and coding process.
A total of 3156 studies were screened.Of them, 2971 studies were excluded through abstract screening, and 77 duplicates were removed from the remaining 185 studies.Eighty-eight more studies were eliminated by reading the full texts.One study was excluded in the last retrieval because the participants were participating in active resistance training, although their average BMI was >25 kg/m 2 .Nineteen studies entered the final meta-analysis, with 11 studies in the isocaloric condition and the other eight in the ad libitum intake condition.Six studies involved e-TRE, seven studies involved d-TRE, and eight articles did not specify the eating time.

Energy intake and eating window are associated with moderate weight loss
The percentage weight change 42 from baseline to the endpoint between the intervention and control groups was analyzed in 19 studies with 22 arms (Figure 2).TRE effectively reduced body weight percentage compared to the control group (À2.04%, 95% confidence interval (CI): À2.57 to À1.50; low certainty evidence).Meta-regression was conducted to explore the relationship between energy intake, eating window, intervention duration, and weight loss percentage.Energy intake (Z = 5.23, p < 0.001) and eating window (Z = 2.26, p = 0.024) were significantly associated with weight loss percentage but not the intervention duration length (Z = 0.36, p = 0.72).No significant publication bias was found (Egger's test p = 0.93).
The forest plot represents the effect of TRE on weight change percentage compared to the non-TRE group.The number of participants (Total), the mean difference (Mean), and the standard deviations (SD) of the primary outcomes from baseline to the endpoint for the intervention and control groups were used to calculate the mean difference and 95% confidence interval of weight change in the included studies.The random effects model was used to estimate the pooled effect size and account for possible heterogeneity.

TRE reduces energy intake intentionally or unintentionally
The change in energy intake from baseline to the endpoint between the intervention and control groups was analyzed in 13 studies with 16 arms (Figure 3).Individuals in the TRE group reduced energy intake by 201.77 kcal/day intentionally or unintentionally compared to the control group (À201.77kcal, 95%CI: À304.12 to À99.43; low certainty evidence).Meta-regression found that the length of the eating window (Z = 2.05, p = 0.040) significantly affected energy intake but not the intervention duration (Z = 0.57, p = 0.57).There was no significant publication bias in the reporting of energy intake (Egger's test p = 0.92).
The forest plot represents the effect of TRE on actual energy intake compared to the non-TRE group.The number of participants (Total), the mean difference (Mean), and the standard deviations (SD) of the primary outcomes from baseline to the endpoint for the intervention and control groups were used to calculate the mean difference and 95% confidence interval of the actual intake in the included studies.The random effects model was used to estimate the pooled effect size and account for possible heterogeneity.

TRE, with or without energy prescription, improves systolic blood pressure and glycemic levels
In the metabolic risk factors (Table 3) analyzed, TRE, with or without energy prescription, effectively improved systolic blood pressure (À2.42%, 95%CI: À4.34 to À0.50, low certainty evidence), fasting glucose levels (À2.57mg/dL, 95%CI: À4.73 to À0.42, very low certainty evidence), fasting insulin levels (À1.81 mIU/mL, 95%CI: À3.24 to À0.38, very low certainty evidence), and HbA1c levels (À0.34, 95%CI: À0.60 to À0.07, very low certainty evidence) but did not benefit other metabolic parameters compared to the control group.a One study did not report gender specifically, 20 so we excluded this study when calculating the gender proportion.These studies only reported the total number of females in their samples.Thus, the number of females in the intervention and control groups was not available.Data on study characteristics, participants, dietary regimes, measurement of energy intake, and primary and secondary outcomes were extracted from the included studies.The macronutrient, consultation provided, physical exercise intensity, and methodology for measuring adherence were also extracted because they are vital factors for conducting a successful TRE intervention. 37For detailed information, see Table S3.a This study contained two intervention groups and one control group, thus was coded and extracted as two separate trials with the same control group in this table.

Energy reduction brings more benefits than isocaloric diets in TRE strategies
To test the crucial effect of actual energy consumption on the TRE diet, we classified the TRE regime as an isocaloric condition (energy prescribed) versus an ad libitum intake condition (energy not prescribed) to conduct a subgroup analysis (Table 4).In the analysis of weight loss percentage and energy intake, individuals lost 3.08% body weight (À3.08%, 95%CI: À3.42 to À2.73) compared to the control group under the ad libitum condition and showed a lower percentage of weight loss under the isocaloric condition (À1.25%, 95%CI: À1.74 to À0.75).
Compared to the considerable energy reduction between the TRE and control groups under the ad libitum condition (À445.96kcal, 95% CI: À468.58 to À423.33), there was no significant energy reduction in the TRE group in the isocaloric condition (À51.53 kcal, 95%CI: À106.62 to 3.55).Taken together, we attributed the weight loss under the isocaloric condition to the benefits of alignment with the eating time of day since the energy intake was not different in this condition.Yet the fact that considerable energy deficit (À445.96kcal/d) and relatively modest body weight loss (À3.08%) in the ad libitum condition was confusing.We then used the Body Weight Planner (niddk.nih.gov/bwp) in NIH (National Institute of Health) to compare the anticipated and actual weight loss given the energy deficit participants reported. 44esults demonstrated that the actual weight loss led by this reported energy deficit was less than predicted, indicating that participants may reported excessive daily energy reduction.
In conclusion, TRE demonstrated improvements in weight loss, anthropometric parameters, body composition, and metabolic health measurements.Subgroup analysis results based on energy prescription support the hypothesis that energy restriction contributed to these benefits, as body weight and metabolic parameters improved less once energy intake was controlled.

Eating time of day also improves the health status of patients on TRE diets
The subgroup analysis described above demonstrated the role of energy deficits in improving health outcomes in TRE regimes, yet improvements in weight loss percentage, anthropometric parameters, body composition, and some metabolic indicators still existed under the isocaloric intake condition.Since many studies have confirmed the significance of eating time of day in the TRE diet, we conducted further subgroup analysis by taking the time of day into consideration, categorizing four subtypes as e-TRE and d-TRE, with and without energy prescription, to separate the effect of eating time of day from energy consumption (Table 5).We investigated whether the alignment with eating time of day could lead to benefits under the ad libitum condition (energy not prescribed) and, if so, whether these benefits still exist once the amount of energy was controlled (energy prescribed).
Weight loss was greater in e-TRE than in d-TRE under the ad libitum condition, and decreased in e-TRE but not in d-TRE under the isocaloric condition.Energy intake reduced more in e-TRE than in d-TRE under the ad libitum condition but did not differ under the isocaloric condition.Waist circumference, fat mass, lean body mass, fasting insulin levels, and HbA1c levels showed greater improvements in e-TRE than in d-TRE without energy prescription but did not differ once the amount of energy was controlled.HOMA-IR improved only in e-TRE when the energy amount was not prescribed.These results demonstrated that for those health outcomes that improved under the ad libitum condition, the alignment with the time of day could, to some extent, strengthen the utility of TRE.

DISCUSSION
The current systematic review and meta-analysis explored the effect of TRE on weight loss and metabolic health in individuals with overweight and obesity statuses based on randomized controlled trials involving 19 studies with 22 intervention arms and 1201 participants.To our knowledge, this was the first systematic review and meta-analysis investigating the reason why TRE leads to weight loss and improves metabolism.The data indicated that TRE was effective in weight loss and metabolic health, and these benefits were due to a joint effect of energy restriction and eating time of day.Subgroup analysis of four subgroups further indicated that energy restriction plays a dominant role compared to the eating time of day.In the isocaloric condition, only e-TRE improved health outcomes while d-TRE with ad libitum intake was also effective, with considerable energy reduction, it was not as beneficial as e-TRE.

Weight loss and energy intake in TRE strategies
TRE reduced body weight and actual energy consumption but did not achieve a clinical significance of 5% weight loss.6]45 Meta-regression showed that weight loss was significantly related to energy consumption (Z = 5.23, p < 0.001) and the eating window (Z = 2.26, p = 0.024) but was not correlated with the intervention duration (Z = 0.36, p = 0.72).This demonstrated an increased capacity for TRE to cause weight loss by a greater calorie deficit and shorter eating window within a safe range. 46Restricting eating window can reduce eating window intentionally or unintentionally and achieve weight loss.Since most people have a daily eating window that exceeds 12 h, 19,[47][48][49] previous studies reported 8 h to be a safe and adherable eating period, and an excessively restricted eating window might ironically result in an increased risk of binge eating.The absence of a correlation between weight loss and intervention duration suggested that more studies with longer and diverse intervention durations are needed to explore the ambiguous effects of intervention duration on TRE outcomes.Some research suggested that the threshold for the effect of TRE strategies was around 12 weeks when the intervention effect was likely to be the best with the highest dietary adherence, after which the effectiveness might decrease as the adherence rate lowered. 11,50 43For detailed information, see Table S6.For risk of bias assessment see Tables S4 and S5.
Therefore, the relevance of energy balance, the eating window, and body weight need to be rigorously assessed, and more objective measurements of daily energy intake and compliance are needed. 51Although the weight loss magnitude was modest, subgroup analysis indeed showed the advantage of eating time-of-day in TRE.For people with metabolic symptoms, TRE can also be used as a strategy to improve their metabolic indicators.Subgroup analysis indicated more weight loss under the ad libitum intake condition than the isocaloric condition.Moreover, both e-TRE and d-TRE reduced body weight under the ad libitum condition with substantial energy deficit (approximately $450 kcal/d), but e-TRE was more effective, and the weight loss benefits of d-TRE disappeared when the energy intake was actually the same (isocaloric condition).The weight loss of e-TRE under prescribed conditions suggested that, to some extent, TRE could indeed improve the health status of people with overweight and obesity statuses through appropriate time-of-day eating in the absence of energy reductions.In conclusion, it is possible that weight loss in the TRE strategy might have been driven by the combination of energy deficit and eating time of day.However, energy reduction had a more robust effect.

TRE improves anthropometric parameters, body composition, and some metabolic outcomes
In the analysis of anthropometric parameters and body composition, TRE reduced waist circumference, BMI, fat mass, and lean body mass but did not affect visceral fat.From this perspective, TRE may not improve fat mass loss while maintaining lean body mass compared to traditional calorie-restricted diets, where weight loss is always accompanied by a concomitant reduction in lean body mass. 4,48,52The decrease in lean mass may have been due to the inclusion of individuals with overweight and obesity statuses in this meta-analysis, and physical exercise with adequate protein intake can regulate the loss of lean tissue. 53The different changes in lean body mass under isocaloric and ad libitum conditions in the subgroup analysis illustrate the importance of energy restriction in sustaining lean body mass.
In metabolic risk factors, TRE improved systolic blood pressure and glycemic (fasting glucose, fasting insulin, and HbA1c levels) indicators, while other parameters, including diastolic blood pressure, triglyceride, total cholesterol, high-density cholesterol, low-density cholesterol levels, HOMA-IR, and resting metabolic rate did not change between the TRE and control groups.Blood pressure is related to metabolic disease and heart failure, but the results of studies on the impact of TRE on blood pressure have been mixed.This meta-analysis found decreased systolic blood pressure in the TRE group under the ad libitum intake condition with sizable energy reduction, whereas diastolic blood pressure remained the same, consistent with another recently published meta-analysis. 45,54A study in shift workers found a significant decrease in blood pressure among participants with elevated systolic blood pressure (R130 mmHg) or diastolic blood pressure (R85 mmHg) compared to those with normal initial levels, indicating the salience of TRE in people with elevated cardiometabolic risks at baseline. 55In addition, subgroup analysis showed that systolic blood pressure was decreased under the ad libitum condition and that e-TRE was more effective compared to d-TRE only with relatively lower energy intake.
Plasma lipids are another factor affecting metabolic disease as well as cardiovascular disease.However, the benefits of TRE are more salient in people with metabolic syndrome, 39 resulting in non-significant improvements in triglycerides, high-density cholesterol, and lowdensity cholesterol levels in this meta-analysis where most of the participants were not patients with metabolic symptoms.Besides improved triglycerides under the ad libitum condition, lipid levels between the isocaloric and ad libitum subgroups were not different.In contrast to the utility of TRE in people with metabolic syndrome, TRE showed no improvement in lipid levels in this meta-analysis, suggesting that TRE may act more as anti-hypertensive therapy in those with elevated baseline lipid levels. 56Glycemic levels and insulin resistance are factors that contribute to both metabolic disease and cardiovascular disease.Previous clinical data and this meta-analysis demonstrated a robust effect of TRE on fasting glucose and fasting insulin levels, indicating TRE as a useful treatment for type 2 diabetes, where glucose management is critical to minimizing diabetes-associated complications and improving health and the quality of life. 57,58onsistent with the conclusions of some previous studies that e-TRE was superior to d-TRE in terms of metabolic improvement, this metaanalysis found similar results that body composition and metabolic parameters in e-TRE were better than those in d-TRE under the ad libitum intake condition, while d-TRE showed no benefits once energy intake was controlled.In other words, the advantages of energy deficits for weight loss and metabolism were more salient than those of the eating time of day in TRE strategies.Individuals who consumed energy amounts over baseline levels could gain weight even using the TRE strategy characterized by the time of day.Poor diet quality, such as low-nutrient, high-fat foods, could also limit the effectiveness of TRE. 12 The estimate of most outcomes in this meta-analysis did not reach clinical significance.However, the intervention effect of TRE in health improvement was greatest in the e-TRE subgroup with considerable energy deficit.The relatively modest magnitude has been demonstrated in some previous meta-analysis, some attributed this to the short intervention duration 45 while other did not find the association between health improvement and duration. 34Thus, more long-term TRE studies are needed to get a more solid conclusion, both in people with overweight or obesity status and patients with metabolic symptoms.
In conclusion, TRE could effectively lead to modest weight loss, decreases in waist circumference, body mass index, fat mass, and lean body mass, and improvements in systolic blood pressure, fasting glucose, fasting insulin, and HbA1c levels relative to the control group.Yet, TRE did not impact visceral fat, diastolic blood pressure, triglyceride, total cholesterol, high-density cholesterol, low-density cholesterol levels, HOMA-IR, or resting metabolic rate.Subgroup analysis based on energy intake and eating time of day suggested that energy restriction and eating time of day collectively led to weight loss and improved metabolic health in the TRE diet strategy, but the effect of energy reduction was more vital.

Limitations of the study and future TRE research
A limitation of this systematic review and meta-analysis was that several related outcomes were not examined due to a small number of trials, and the intervention strategies were relatively simple.Except for one study by Liu et al. 52 with an intervention duration of one year (48 weeks), the remaining studies had relatively short intervention durations.Most were 8 and 12 weeks and lacked follow-up on the long-term effects of TRE strategies.Some studies 11,51 suggested that the intervention effect of TRE showed an inverted U-shaped curve that increased initially and then decreased with the intervention duration.However, in this meta-analysis, we were not able to perform a nonlinear meta-regression to test this hypothesis due to the limitations of the intervention duration lengths in the included studies.Second, subgroup analysis and meta-regression using group-level data suffer from ecological fallacy.Therefore, the conclusion does not lead to a causal relationship and cannot be generalized to the individual level. 59,60That is, the intervention strategies demonstrated to be effective in this meta-analysis are not necessarily helpful to a certain person with overweight or obesity status.Moreover, given that there was only one study, that by Zhang et al., 21 under e-TRE without energy prescription in the subgroup analysis, great caution must be taken when describing the interaction of energy consumption and eating time of day on TRE for weight loss and metabolic improvement. 61Third, we did not distinguish patients with metabolic symptoms from other participants as the number of RCT studies in those patients was small, yet previous studies reported TRE to be a more efficient dietary strategy for people with metabolic syndrome. 56Forth, we did not include Ramadan fasting in this metaanalysis because the restriction of eating time-of-day after sunset and at night did not apply to the majority of non-Muslim population.Given the relative youth of the body of literature examining TRE in humans, future studies should attempt to identify and empirically test the influence of multiple different intervention approaches on TRE for weight loss and improvements in metabolic parameters, including what, when, and how much individuals eat daily.Fifth, the adequacy of energy intake measures was not considered in the exclusion criteria when screening the articles.Although most clinical RCTs measured adherence during TRE, only four of them reported adherence quantitatively as the percentage of the participating days or the number of adherent days per week.Moreover, self-reported measures of adherence, such as remote video by Skype or daily adherence logs, rely on participants' honesty and may be affected by social desirability bias. 11,62In addition to using adherence to predict long-term weight maintenance, the distribution of protein and carbohydrate intake is crucial for understanding the full benefit of TRE as these macronutrients are vital to modifying body composition and glycemic levels. 63,64Biological hunger and satiety can result in urges to eat and are related to TRE eating time.However, few studies have measured leptin levels or reported subjective hunger.A recent study proposed that participants consuming a morning-loaded diet reported significantly lower hunger and reduced caloric intake at lunch and evening, thus contributing to enhanced weight loss. 65Additional research is needed to examine the fluctuation of hunger levels throughout the day and its association with eating window time.Gender is also a physiological factor that could affect TRE outcomes.Most participants in this meta-analysis were female (65.07%), whereas male participants accounted for a large proportion only in studies of individuals participating in physical activities or special populations, such as firefighters or athletes.Thus, more studies are needed to examine whether gender differences have an impact on TRE strategy with the same baseline body composition and metabolic risk factors.Some public social issues should also be considered when promoting the TRE strategy to a broader range of people.The disrupted activity-rest cycle caused by artificial light and industrialization indirectly disrupts the natural daily cycle of feeding and fasting and facilitates excessive caloric intake. 7Additionally, it is difficult for shift workers, people with low economic levels, and those who lack social support to restrain their eating window to less than 12 h a day. 43,66Therefore, the scope of future studies should be expanded to understand the feasibility of TRE in these populations.

Figure 1 .
Figure 1.Scheme of study selection

Figure 2 .
Figure 2. Meta-analysis results of weight loss

Figure 3 .
Figure 3. Meta-analysis results of energy intake

Table 1 .
Participant information

Table 2 .
Basic characteristics of the included studies

Table 3 .
Results of meta-analysis of outcomes WL, weight loss; EI, energy intake; WC, waist circumference; BMI, body mass index; FM, fat mass; BLM, lean body mass; VF, visceral fat; SBP, systolic blood pressure; DBP, diastolic blood pressure; HR, heart rate; TG, triglycerides; TC, total cholesterol; HDL, high-density cholesterol; LDL, low-density cholesterol; FG, fasting glucose; FI, fasting insulin; HOMA-IR, Homeostasis Model Assessment-Insulin Resistance; HbA1c, Hemoglobin A1c; RMR, resting metabolic rate.a The Grading of Recommendations Assessment; Development and Evaluation (GRADE) was assessed according to the Cochrane Handbook, the GRADE handbook and other articles.

Table 5 .
The results of subgroup analysis based on energy prescription and eating time of day