Effects of timing and eating duration of time restricted eating on metabolic outcomes: systematic review and network meta-analysis

Abstract

Objectives

To evaluate the effect of overall time restricted eating on metabolic health outcomes, and to identify the optimal types of time restricted eating in terms of timing and duration of eating.

Design

Systematic review and network meta-analysis.

Data sources

PubMed, Embase, and the Cochrane databases, from inception to 3 January 2023.

Eligibility criteria for selecting studies

Randomised controlled trials investigating the relation between time restricted eating (intervention period >one month) and metabolic health outcomes in humans.

Results

41 randomised controlled trials of 2287 participants were included. Time restricted eating was categorised according to time of eating (early, mid-, late, and self-selected; last meal eaten before 1700, between 1700 and 1900, after 1900, or chosen by participant, respectively) and specific duration or window of eating each day (<8 hours, eight hours, and >8 hours). Compared with usual diets, overall time restricted eating significantly improved body weight, body mass index, fat mass, waist circumference, systolic blood pressure, and levels of fasting blood glucose, fasting insulin, and triglycerides. For different time restricted eating subtypes, early time restricted eating consistently showed higher P score rankings for anthropometric measurements (P score range 0.62-0.86, except for fat free mass-lean mass; a score closer to one indicating more favourable subtype) and glycaemic parameters (P score range 0.66-0.99). Compared with late time restricted eating, early time restricted eating significantly reduced body weight (mean difference −1.15 kg, 95% confidence interval −1.86 to −0.45) and fasting insulin concentrations (−3.32 μIU/ml, −5.36 to −1.28; 1 μIU/mL=6.95 pmol/L) and the certainty of the evidence was high. P value rankings for eating duration were inconsistent. Assessment of risk of bias, based on the risk of bias 2 tool, found that most of the included studies (90%) were rated as low risk of bias. In the confidence in network meta-analysis (CINeMA) assessment, about 60.2% of the network evidence showed moderate to high certainty. Inconsistency was generally low (I²<75% for 87% of associations).

Conclusions

Time restricted eating overall improved metabolic health outcomes compared with usual diets, and early time restricted eating was superior to late time restricted eating. The association between duration of eating and metabolic health outcomes was inconsistent.

Systematic review registration

PROSPERO CRD42022302737.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Cardiovascular disease is a leading cause of death, with modifiable metabolic risk factors having a major role

• Time restricted eating improves metabolic health without requiring a reduction in overall energy intake

• The specific effects of different eating times and durations within this eating regimen have not been thoroughly investigated

WHAT THIS STUDY ADDS

• Compared to a usual diet, time restricted eating was associated with significant improvements in body weight, body mass index, fat mass, waist circumference, systolic blood pressure, and levels of fasting blood glucose, fasting insulin, and triglycerides

• Early and mid-time restricted eating in general had better efficacy than late time restricted eating, particularly for anthropometric measurements and glycaemic parameters

• Late time restricted eating, especially when combined with a longer eating duration, was generally least effective for most metabolic outcomes

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE, OR POLICY

• Early or mid-time restricted eating could be prioritised for better metabolic health outcomes

• Late time restricted eating with extended eating durations seems less effective, informing future dietary strategies and research

Introduction

Non-communicable diseases constitute more than half of the global disease burden.¹ For deaths related to non-communicable diseases, cardiovascular diseases are responsible for about 50%.² Modifiable metabolic risk factors, including high body mass index,^{3 4} high blood pressure,³ and abnormal glycaemic³ and lipid profiles,³ account for a substantial proportion of cardiovascular deaths, indicating the importance of optimising metabolic health.^{5 6}

Dietary strategies aimed at improving metabolic health have mainly focused on the content of diets⁷,⁹ or continuous energy restriction.¹⁰ Recent insights, however, have highlighted the major impact of the eating-fasting cycle on various physiological and metabolic processes.¹¹ Intermittent fasting, a dietary approach based on regulating the timing of food intake rather than focusing only on content,¹² has gained much attention.¹³ Intermittent fasting is an alternative approach to improving metabolic health that may be more feasible for sustained long term adherence.¹⁴

Typical types of intermittent fasting include alternate day fasting, periodic fasting, and time restricted eating.¹⁵ Time restricted eating is a distinctive approach to intermittent fasting; food intake is restricted to a specific duration or window of time each day (typically <8-12 hours/day), instead of specifying fasting days, and reducing overall energy intake is not required.^{12 16} Controlled feeding trials with isocaloric time restricted eating have shown improvements in metabolic parameters, even without concurrent weight loss.¹⁷ This finding indicates that time restricted eating may provide some metabolic benefits through pathways unrelated to intermittent energy reduction.

Optimal time restricted eating not only emphasises the restriction of the food intake window but should also align meal times with the body's natural circadian rhythm.¹⁸ Previous research mainly compared time restricted eating with usual diets or other dietary strategies, such as continuous energy restriction. The differential effects of different eating times and durations of eating within the time restricted eating approach were not studied, which are crucial in informing optimal types of time restricted eating for maximum benefits to metabolic health. Excessive food intake at night can cause circadian misalignment between the central and peripheral clocks.¹⁹ Thus time restricted eating started during the early part of the day (early time restricted eating) is generally considered to be superior. Limited direct comparisons of early versus late time restricted eating have been conflicting, however, impeding firm conclusions.²⁰,²² To look at this knowledge gap, we conducted a network meta-analysis incorporating direct and indirect evidence. We first evaluated the metabolic effects of overall time restricted eating, and subsequently determined which subtypes of time restricted eating, based on timing and duration of eating, gave the best outcomes for metabolic health.

Methods

Literature search

We conducted a comprehensive search of PubMed, Embase, and the Cochrane database, from inception to 3 January 2023. We used a combination of keywords related to intervention (synonyms of time restricted eating) and outcomes associated with metabolic health (online supplemental table S1). We included randomised controlled trials examining the association between time restricted eating and metabolic health outcomes in humans. We excluded literature that had an intervention duration of <1 month.

Study selection and data extraction

Two reviewers independently screened 85% of the titles and abstracts (κ=0.82) and 85% of the full text articles (κ=0.95), with substantial agreement between the reviewers. The remaining 15% were screened by one reviewer with the same eligibility criteria. If disagreements could not be resolved through consensus, a third researcher (L-WC) acted as an arbitrator. Data extraction was independently performed by two reviewers for 85% of the studies. Inter-reviewer agreement was high (concordance rate 94%), indicating substantial consistency.²³ Given this high concordance, one reviewer extracted data for the remaining 15% of the articles. The information extracted was first author, publication year, number of participants, population characteristics, intervention duration, and the timing of the time restricted eating window. We also collected effect estimates for metabolic health outcomes, specifically the changes in values between and after interventions.

Outcomes of interest

Outcomes were anthropometric measures (body weight, body mass index, fat mass, fat free mass-lean mass, and waist circumference), blood pressure measurements (systolic and diastolic blood pressure), glycaemic parameters (fasting blood glucose, fasting insulin, and haemoglobin A_1c (HbA_1c) levels, and homeostatic model assessment for insulin resistance (HOMA-IR)), and lipid profiles (levels of triglycerides, total cholesterol, low density lipoprotein cholesterol, and high density lipoprotein cholesterol).

Dietary patterns

We categorised time restricted eating according to the timing of the last meal. For early time restricted eating,¹³ the last meal was eaten before 1700, for mid-time restricted eating,¹³ the last meal ended between 1700 and 1900, and for late time restricted eating, eating was concluded after 1900.¹³ Self-selected time restricted eating referred to studies that allowed participants to choose their own eating time.

We also classified diets based on a specific duration or window of eating each day: <8 hours (ie, eating duration <8 hours/day), eight hours, and >8 hours. Furthermore, because of inconsistent results for the duration of time restricted eating, we conducted further analyses by combining the early, mid-, and late categories with duration of eating. We subdivided time restricted eating into the groups early and mid-time restricted eating ≤8 hours, early and mid-time restricted eating >8 hours, late time restricted eating ≤8 hours, and late time restricted eating >8 hours.

Dietary regimens of continuous energy restriction, irrespective of additional dietary guidance, were categorised as continuous energy restriction. Those offering dietary advice without energy restrictions were labelled as dietary advice. For intermittent fasting regimens other than time restricted eating, the original classifications in the primary literature were maintained without further categorisation. Participants who were not advised to make any changes were categorised as adhering to their usual diet (online supplemental table S2).

Risk of bias and quality of evidence

We used the Cochrane risk of bias tool (RoB 2.0) to evaluate the risk of bias in the included studies.²⁴ This comprehensive grade was derived by evaluating five domains: bias arising from the randomisation process; bias caused by deviations from the intended interventions; bias owing to missing outcome data; bias in outcome measurement; and bias in the selection of the reported result. Studies were then categorised as having an overall risk of bias, graded as high, some, or low risk of bias.

We used the confidence in network meta-analysis (CINeMA) framework to determine the confidence level of the network meta-analysis. This assessment considered six domains: bias within studies, reporting bias (online supplemental figures S1 and S2), indirectness, imprecision, heterogeneity, and incoherence. We then assigned an overall confidence rating of very low, low, moderate, or high certainty of evidence (online supplemental text S1).²⁵

Statistical methods

We extracted mean changes, standard deviations, and sample sizes from the included studies, converting 95% confidence intervals (CIs), standard errors, or interquartile ranges into standard deviations based on methods described in the Cochrane handbook.²⁶ We first evaluated the effects of overall time restricted eating on various metabolic outcomes. The primary analyses focused on the effects of different time restricted eating subtypes, including early, mid-, and late time restricted eating, as well as variations in the duration of the eating window. For each outcome, we conducted a separate random effects network meta-analysis with the netmeta package of R version 4.2.3,²⁷ which implemented a frequentist contrast based model. For the primary outcomes, we performed pairwise comparisons between dietary interventions and assessed significance based on whether the 95% confidence interval crossed the null value. We also calculated treatment rank probabilities with the P score, ranging from 0 to 1, with a higher score indicating a more effective treatment.²⁸

To evaluate the transitivity assumption, the baseline distribution of body weight, systolic blood pressure, fasting blood glucose levels, and total cholesterol levels was assessed for all treatment comparisons. Coherence, defined as the agreement between direct and indirect evidence, was assessed with the design-by-treatment interaction model and the node splitting method. A P value <0.05 for testing the difference in the estimates given by direct and indirect evidence indicated incoherence.^{29 30} Heterogeneity was assessed as variance between studies (τ²). Inconsistency was assessed with the I² statistic, which describes the percentage of total variation across studies that is caused by heterogeneity rather than chance. If I² was >75%,²⁷ we performed meta-regression analyses to explore potential effect modifiers, including baseline body mass index, age, intervention duration, disease status, and whether the time restricted eating group had energy restrictions. Likelihood ratio tests were used to test interaction effects. We performed sensitivity analyses for healthy populations and interventions lasting 1-6 months.

Effects on metabolic outcomes

To summarise our network meta-analysis evidence, we first categorised the certainty of the results, based on the CINeMA assessment, into moderate to high certainty and very low to low certainty. We then examined the effects of dietary interventions on metabolic health. A major effect on metabolic health was defined as a dietary intervention that was significantly better than the usual diet (meaning that the 95% confidence interval did not include the null value) as well as significantly better than another dietary strategy that was also significantly better than the usual diet. A minor effect on metabolic health was defined as a dietary intervention that was significantly better than the usual diet but not significantly better than any other dietary strategy. If a dietary intervention was not significantly different from the usual diet, we categorised it as having no effect.³¹ We then combined the categories, resulting in six possible groups: major effect with moderate to high certainty, major effect with low to very low certainty, minor effect with moderate to high certainty, minor effect with low to very low certainty, no effect with moderate to high certainty, and no effect with low to very low certainty. We also compared dietary interventions with CER using the same classification framework to evaluate their relative effects.

Protocol deviations

Based on practical considerations during implementation, minor adjustments were made to the original PROSPERO protocol to enhance methodological rigour and feasibility. Although the original protocol proposed the inclusion of clinical trials with various designs, the final analysis was restricted to randomised controlled trials to improve internal validity and comparability across studies. Ramadan fasting conceptually resembles time restricted eating, but we excluded studies conducted during Ramadan because they involve other concurrent lifestyle modifications that could introduce confounding. Also, several planned subgroup analyses and outcomes, such as metabolic syndrome and type 2 diabetes, were not conducted because of unavailable or insufficient data. We also performed more exploratory analyses that combined both the timing and duration of the eating window (eg, early time restricted eating ≤8 hours v early time restricted eating >8 hours) because of inconsistent findings on the optimal time restricted eating duration. These analyses were not prespecified in the protocol and should therefore be interpreted as hypothesis generating. Online supplemental table S3 summarises all protocol deviations and their justifications.

Patient and public involvement

Because the study was a systematic review of previously published studies, involving patients or the public in the design, or conduct, or reporting, or dissemination plans of our research was not appropriate or possible. We intend to share our results on research platforms after the publication of the manuscript.

Results

Of the 4213 articles identified in PubMed, Embase, and the Cochrane database, 41 randomised controlled trial studies met the inclusion criteria after screening (online supplemental figure S3).

Study characteristics, risk of bias, and certainty of evidence

Forty one randomised controlled trials of 2287 participants were included in our analysis, and 42.0% of participants were women. Data for sex were taken from the studies rather than from patient reported gender. The median of study level mean participant age was 37.2 years (range of means 19.3-69.5 years) and the median of study level mean body mass index was 27.8 (range of means 21.9-39.6). Five studies included participants with metabolic syndrome, type 2 diabetes, pre-diabetes, or non-alcoholic fatty liver disease, and 12 studies imposed restrictions on calorie intake for time restricted eating. The median duration of intervention for all studies was eight weeks (range of means 4-48 weeks) (online supplemental table S4). The classification of time restricted eating based on the timing of eating criteria outlined above resulted in six arms for early time restricted eating,¹⁷²⁰,^{22 32 33} nine arms for mid-time restricted eating,^{1722 34},⁴⁰ 19 arms for late time restricted eating,^{2133 35 40},⁵⁵ and 13 arms for self-selected time restricted eating.⁵⁶,⁶⁸ For time restricted eating duration, 10 arms had an eating window of <8 hours,^{17 22 36 37 39 40 50 52 53 68} 24 arms an eight hour window,^{2021 32},^{34 40} and 10 arms a window of >8 hours.^{17 35 38 54 56 57 62 63 66 67} Online supplemental table S5 shows the data extracted from all of the included studies. Online supplemental figures S4-S6 present pairwise network plots for all studies.

The design-by-treatment interaction model (online supplemental tables S6 and S7) and node splitting analysis (online supplemental tables S8 and S9) indicated that most network evidence was coherent with the direct evidence; the assumption of transitivity was not seriously violated (online supplemental figures S7 and S8). Most of the studies (90.2%) were rated as low risk of bias (online supplemental figure S9) based on the risk of bias 2 tool. CINeMA assessment of the network evidence for early, mid-, late, and self-selected time restricted eating was 29.8% high certainty, 32.5% moderate certainty, 35.2% low certainty, and 2.4% very low certainty of evidence (online supplemental tables S10 and S11). Similarly, for time restricted eating duration, the distributions were 24.8% high certainty, 32.7% moderate certainty, 38.2% low certainty, and 4.2% very low certainty of evidence.

Time restricted eating versus other dietary strategies and metabolic health

Anthropometric measures

Compared with usual diets, overall time restricted eating was associated with significant improvements in several anthropometric indicators. Specifically, time restricted eating reduced body weight (mean difference −2.15 kg, 95% CI −2.56 to −1.74), body mass index (−0.76, −1.11 to −0.41), fat mass (−1.32 kg, −1.61 to −1.03), and waist circumference (−1.63 cm, −3.02 to −0.23). Also, time restricted eating significantly reduced body weight compared with dietary advice (−0.84 kg, 95% CI −1.56 to −0.13). These improvements, however, were accompanied by a reduction in fat free mass-lean mass (−0.84 kg, 95% CI −1.27 to −0.41). Although time restricted eating reduced fat mass compared with usual diets, it was associated with an increase of 0.96 kg in fat mass when compared with alternate day fasting (95% CI 0.08 to 1.85) (figure 1).

Figure 1. Effects of time restricted eating (TRE) compared with other dietary strategies on metabolic health outcomes. Other dietary strategies were alternative day fasting (ADF), week on week off (WOWO), continuous energy restriction (CER), dietary advice, and usual diet. Metabolic health outcomes measured were anthropometric measures (body weight, body mass index, fat mass, fat free mass-lean mass, waist circumference), blood pressure measurements (systolic blood pressure, diastolic blood pressure), glycaemic parameters (fasting blood glucose, fasting insulin, and haemoglobin A_1c (HbA_1c) levels, and homeostatic model assessment for insulin resistance (HOMA-IR)), and lipid profiles (levels of triglycerides, total cholesterol, low density lipoprotein cholesterol, and high density lipoprotein cholesterol) (1 mg/dL=0.06 mmol/L), fasting insulin concentrations (1 μIU/mL=6.95 pmol/L), haemoglobin A_1c concentrations (HbA_1c; HbA_1c 1%= (1×10.93) − 23.50 mmol/mol), homeostatic model assessment for insulin resistance (HOMA-IR), and concentrations of triglycerides (1 mg/dL=0.01 mmol/L), total cholesterol (1 mg/dL=0.03 mmol/L), low density lipoprotein cholesterol (1 mg/dL=0.03 mmol/L), and high density lipoprotein cholesterol (1 mg/dL=0.03 mmol/L). Effect estimates are mean difference (95% confidence interval (CI)).

Blood pressure

Compared with usual diets, time restricted eating significantly reduced systolic blood pressure by 4.94 mm Hg (95% CI −6.92 to −2.95), but had no effect on diastolic blood pressure (figure 1).

Glycaemic parameters

Compared with usual diets, time restricted eating was associated with significant reductions in concentrations of fasting blood glucose (−3.69 mg/dL, 95% CI −6.19 to −1.18; 1 mg/dL=0.06 mmol/L) and fasting insulin (−0.56 μIU/mL, −0.89 to −0.23; 1 μIU/mL=6.95 pmol/L). We found no significant differences for HbA_1c or HOMA-IR (figure 1).

Lipid profile

Compared with usual diets, time restricted eating was associated with a significant reduction in triglyceride levels (−10.12 mg/dL, 95% CI −19.09 to −1.15; 1 mg/dL=0.01 mmol/L). However, when compared with alternate day fasting, time restricted eating was associated with an increase of 19.79 mg/dL in total cholesterol levels (95% CI 5.18 to 34.4; 1 mg/dL=0.03 mmol/L) (figure 1).

Subtypes of time restricted eating and metabolic health

Anthropometric measures

Compared with late time restricted eating, early time restricted eating resulted in a significant reduction in body weight of 1.15 kg (95% CI −1.86 to −0.45, high certainty of evidence) and a reduction in waist circumference of 3.19 cm (−5.54 to −0.84, low certainty). Mid-time restricted eating, compared with late time restricted eating, significantly reduced body weight by 0.94 kg (95% CI −1.63 to −0.24, high certainty), body mass index by 0.48 (−0.94 to −0.02, moderate certainty), and waist circumference by 2.56 cm (−5.11 to −0.01, moderate certainty). Early time restricted eating, however, significantly reduced fat free mass-lean mass by 0.79 kg (95% CI −1.35 to −0.23, high certainty) compared with mid-time restricted eating (online supplemental table S12). According to the P score analysis, early time restricted eating and mid-time restricted eating consistently ranked first or second for most anthropometric measures, except for fat free mass-lean mass (figure 2 and online supplemental figure S10A).

Figure 2. Ranking by P score for early time restricted eating (eTRE), mid-time restricted eating (mTRE), late time restricted eating (lTRE), and self-selected time restricted eating (sTRE). Anthropometric measures are body weight (BW), body mass index (BMI), fat mass (FM), fat free mass-lean mass (FFM-LM), and waist circumference (WC). Blood pressure measurements are systolic blood pressure (SBP) and diastolic blood pressure (DBP). Glycaemic parameters are fasting blood glucose (FBG) levels, fasting insulin (FI) levels, haemoglobin A_1c (HbA_1c) levels, and homeostatic model assessment for insulin resistance (HOMA-IR). Lipid profile includes levels of triglycerides (TG), total cholesterol (TC), low density lipoprotein cholesterol (LDL-C), and high density lipoprotein cholesterol (HDL-C). Effect estimates are P score.

Blood pressure

We found no significant differences in blood pressure outcomes for all types of time restricted eating (online supplemental table S12F and S12G). P score rankings for blood pressure were mixed (figure 2).

Glycaemic parameters

Early time restricted eating significantly reduced fasting insulin concentrations compared with mid-time restricted eating (−2.75 μIU/ml, 95% CI −4.76 to −0.73, high certainty of evidence), late time restricted eating (−3.32 μIU/ml, −5.36 to −1.28, high certainty), and self-selected time restricted eating (−3.69 μIU/ml, −6.05 to −1.33, high certainty) (online supplemental table S12I). Based on the P score analysis, early time restricted eating consistently ranked first or second (figure 2 and online supplemental figure S10A).

Lipid profile

Compared with late time restricted eating, mid-time restricted eating unexpectedly showed a significant decrease in high density lipoprotein cholesterol concentrations (-2.75 mg/dL, 95% CI -5.29 to -0.20 (1 mg/dL=0.03 mmol/L); moderate certainty of evidence) (online supplemental table S12O). In terms of rankings by P score, self-selected time restricted eating was ranked first for reducing levels of triglycerides, total cholesterol, and low density lipoprotein cholesterol (figure 2 and online supplemental figure S10A). Compared with other types of time restricted eating, however, self-selected time restricted eating showed no significant differences for levels of triglycerides, total cholesterol, and low density lipoprotein cholesterol (online supplemental table S12 panel L-N).

Subtypes of time restricted eating versus usual diet and metabolic health

Anthropometric measures

Compared with usual diets, early time restricted eating and mid-time restricted eating were associated with major or minor effects for several anthropometric outcomes. Specifically, early time restricted eating significantly reduced body weight by 2.48 kg (95% CI −3.03 to −1.93; high certainty of evidence), body mass index by 1.00 (−1.41 to −0.58; high certainty), fat mass by 1.35 kg (−1.75 to −0.95; high certainty), and waist circumference by 3.52 cm (−5.74 to −1.30; high certainty). Similarly, mid-time restricted eating was associated with reductions of 2.26 kg in body weight (95% CI −2.66 to −1.86; moderate certainty), 1.06 in body mass index (−1.36 to −0.75; moderate certainty), 1.79 kg in fat mass (−2.39 to −1.20; high certainty), and 2.90 cm in waist circumference (−5.25 to −0.54; moderate certainty). Late time restricted eating showed minor effects, with reductions of 1.32 kg in body weight (95% CI −1.95 to −0.70; high certainty), 0.58 in body mass index (−1.01 to −0.15; moderate certainty), and 1.14 kg in fat mass (−1.58 to −0.70; moderate certainty). Self-selected time restricted eating also showed minor effects, reducing body weight by 1.94 kg (95% CI −2.89 to −0.99; high certainty), body mass index by 0.41 (−0.81 to −0.00, moderate certainty), and fat mass by 1.25 kg (−2.00 to −0.49; moderate certainty). We found significant reductions in fat free mass-lean mass, however, in early time restricted eating (−1.27 kg, 95% CI −1.83 to −0.71; high certainty), late time restricted eating (−1.04 kg, −1.83 to −0.26; high certainty), and self-selected time restricted eating (−1.07 kg, −2.01 to −0.14; moderate certainty) (online supplemental figure S11).

Blood pressure

Compared with usual diets, all subtypes of time restricted eating were associated with minor effects on systolic blood pressure, with reductions ranging from −3.75 to −6.16 mm Hg. We found the largest reduction with early time restricted eating (−6.16 mm Hg, 95% CI −9.20 to −3.12; high certainty of evidence), followed by late time restricted eating (−4.97 mm Hg, −8.33 to −1.60; high certainty), self-selected time restricted eating (−4.27 mm Hg, −8.42 to −0.13; moderate certainty), and mid-time restricted eating (−3.75 mm Hg, −7.42 to −0.08; moderate certainty). None of the time restricted eating subtypes had a significant effect on diastolic blood pressure compared with usual diets (online supplemental figure S11).

Glycaemic parameters

Compared with usual diets, early time restricted eating showed a minor effect in lowering fasting blood glucose concentrations (−6.13 mg/dL, 95% CI −10.51 to −1.76; high certainty of evidence) and a major effect in reducing fasting insulin concentrations (−3.67 μIU/mL, −5.67 to −1.67; high certainty). Mid-time restricted eating also showed a minor effect on fasting insulin concentrations (−0.93 μIU/mL, −1.63 to −0.23; low certainty). Both late time restricted eating and self-selected time restricted eating did not significantly affect any glycaemic parameters compared with usual diets. None of the time restricted eating subtypes showed a significant effect on HbA_1c or HOMA-IR compared with usual diets (online supplemental figure S11).

Lipid profile

Self-selected time restricted eating had a minor effect compared with usual diets, with a significant reduction of 17.03 mg/dL in triglyceride concentrations (95% CI −29.69 to −4.37; moderate certainty of evidence). Late time restricted eating had a minor effect, with a slight increase of 2.15 mg/dL in high density lipoprotein cholesterol concentrations (0.06 to 4.24), but the certainty of the evidence was low (online supplemental figure S11).

We also compared the subtypes of time restricted eating with continuous energy restriction (online supplemental figure S12). In summary, the findings indicated that compared with continuous energy restriction, early time restricted eating significantly reduced body weight (−1.13 kg, 95% CI −1.92 to −0.35; high certainty of evidence) and waist circumference (−2.19 cm, −4.03 to −0.35; moderate certainty), mid-time restricted eating significantly reduced body weight (−0.91 kg, −1.70 to −0.13; high certainty) and fat mass (−0.96 kg, −1.73 to −0.18; high certainty), whereas late time restricted eating significantly increased concentrations of high density lipoprotein cholesterol (3.80 mg/dL, 0.76 to 6.84; moderate certainty).

Eating durations and metabolic health

Anthropometric measures

Among the different time restricted eating durations, only time restricted eating <8 hours significantly reduced waist circumference by 2.65 cm (95% CI −5.14 to −0.17; moderate certainty of evidence) compared with time restricted eating for eight hours (online supplemental table S13E), but the overall P score rankings were inconsistent (figure 3 and online supplemental figure S10B).

Figure 3. Ranking by P score for time restricted eating (TRE) duration: TRE <8 hours, TRE 8 hours, and TRE >8 hours. Anthropometric measures are body weight (BW), body mass index (BMI), fat mass (FM), fat free mass-lean mass (FFM-LM), and waist circumference (WC). Blood pressure measurements are systolic blood pressure (SBP) and diastolic blood pressure (DBP). Glycaemic parameters are fasting blood glucose (FBG) levels, fasting insulin (FI) levels, haemoglobin A_1c (HbA_1c) levels, and homeostatic model assessment for insulin resistance (HOMA-IR). Lipid profile includes levels of triglycerides (TG), total cholesterol (TC), low density lipoprotein cholesterol (LDL-C), and high density lipoprotein cholesterol (HDL-C). Effect estimates are P score.

Blood pressure

We found no significant differences in blood pressure for the different time restricted eating durations (online supplemental table S13F and S13G), and P score rankings showed a mixed pattern (figure 3 and online supplemental figure S10B).

Glycaemic parameters

Time restricted eating duration of <8 hours significantly reduced fasting insulin concentrations by 1.48 μIU/ml (95% CI −2.90 to −0.05; moderate certainty of evidence) compared with a duration of eight hours, and by 1.53 μIU/ml (−2.98 to −0.09; moderate certainty) compared with a duration of >8 hours (online supplemental table S13I). The overall P score rankings, however, were inconsistent for glycaemic parameters (figure 3 and online supplemental figure S10B).

Lipid profile

Time restricted eating duration of <8 hours significantly increased concentrations of total cholesterol by 11.75 mg/dL (95% CI 2.35 to 21.14; high certainty of evidence) compared with a duration of eight hours, and by 15.17 mg/dL (6.73 to 23.62; high certainty) compared with a duration of >8 hours (online supplemental table S13M). In terms of ranking by P score, however, a time restricted eating duration of <8 hours consistently showed a lower ranking than other durations (figure 3 and online supplemental figure S10B).

Time restricted eating duration versus usual diet and metabolic health

Anthropometric measures

Compared with usual diets, time restricted eating of <8 hours showed major and minor effects, resulting in reductions of 2.56 kg in body weight (95% CI −3.25 to −1.86; moderate certainty of evidence), 0.73 in body mass index (−1.29 to −0.18; moderate certainty), 1.70 kg in fat mass (−2.31 to −1.08; high certainty), and 3.18 cm in waist circumference (−5.25 to −1.10; moderate certainty). Time restricted eating duration of <8 hours also reduced fat free mass-lean mass by 1.06 kg (95% CI −1.57 to −0.55; moderate certainty). Time restricted eating duration of eight hours showed major and minor effects, reducing body weight by 2.25 kg (95% CI −2.88 to −1.63; moderate certainty), body mass index by 0.63 (−1.20 to −0.06; moderate certainty), and fat mass by 1.31 kg (−1.70 to −0.93; high certainty). Time restricted eating duration of >8 hours showed a major effect, reducing body weight by 1.99 kg (−2.66 to −1.33; moderate certainty) (online supplemental figure S13).

Blood pressure

All time restricted eating durations had a minor effect on blood pressure, significantly reducing systolic blood pressure by 4.62-6.03 mm Hg compared with usual diets. Specifically, systolic blood pressure was reduced by 6.03 mm Hg (95% CI −9.46 to −2.61; high certainty of evidence) with a time restricted eating duration of <8 hours, by 5.04 mm Hg (−8.94 to −1.13, high certainty) with a duration of eight hours, and by 4.62 mm Hg (−8.82 to −0.42, moderate certainty) with a duration of >8 hours. Diastolic blood pressure was not significantly affected by any of the time restricted eating durations (online supplemental figure S13).

Glycaemic parameters

Time restricted eating duration of <8 hours showed a minor effect on reducing fasting insulin concentrations by 1.98 μIU/ml (95% CI −3.32 to −0.64; high certainty of evidence). Time restricted eating duration of eight hours had a minor effect on fasting blood glucose concentrations, with a reduction of 5.35 mg/dL (95% CI −8.65 to −2.05; high certainty). Time restricted eating duration of >8 hours did not significantly affect any glycaemic parameters compared with usual diets. None of the time restricted eating durations had a significant effect on HbA_1c or HOMA-IR compared with usual diets (online supplemental figure S13).

Lipid profile

Time restricted eating duration of eight hours had a minor effect, reducing triglyceride concentrations by 15.19 mg/dL (95% CI −24.48 to −5.89; moderate certainty of evidence). Time restricted eating duration of >8 hours showed a minor effect, reducing total cholesterol concentrations by 6.54 mg/dL (95% CI −7.51 to −5.58; moderate certainty). Time restricted eating duration <8 hours significantly increased concentrations of total cholesterol by 8.63 mg/dL (95% CI 0.23 to 17.03) and low density lipoprotein cholesterol by 4.54 mg/dL (0.86 to 8.23; 1 mg/dL=0.03 mmol/L), both with high certainty. None of the time restricted eating durations significantly affected levels of high density lipoprotein cholesterol compared with usual diets (online supplemental figure S13).

We also compared the different eating durations with continuous energy restriction (online supplemental figure S14). The findings showed that compared with a continuous energy restriction diet, only time restricted eating duration of <8 hours significantly decreased fat mass (−0.88 kg, 95% CI −1.71 to −0.04; high certainty of evidence) and waist circumference (−2.87 cm, −5.58 to −0.16; moderate certainty), but also increased total cholesterol concentrations (14.47 mg/dL, 3.01 to 25.93; high certainty).

Timing and eating duration combined and metabolic health

Because of the inconsistent effects of time restricted eating duration, we conducted further analyses by integrating the early, mid-, and late time restricted eating categories with eating duration (online supplemental figures S15 and S10C). The results showed that when considering the time restricted eating categories and eating durations simultaneously, early and mid-time restricted eating for a duration of ≤8 hours consistently ranked first or second for P score for anthropometric measures and glycaemic parameters. Late time restricted eating with a duration of >8 hours generally showed the lowest ranking for most of the outcomes. Late time restricted eating for a duration of ≤8 hours showed a different pattern, with a P score ranking of first or second for systolic blood pressure, diastolic blood pressure, fasting insulin levels, HOMA-IR, and levels of total cholesterol, low density lipoprotein cholesterol, and high density lipoprotein cholesterol.

Adherence and adverse events of time restricted eating

Adherence to time restricted eating ranged from 66.3% to 99.0% (online supplemental table S14). In seven studies comparing time restricted eating with continuous energy restriction, adherence was generally higher for time restricted eating. In 14 studies with data, time restricted eating did not cause serious adverse events, but mild adverse events were reported in some participants, including dizziness (n=12) headache (n=6), and constipation (n=7) (online supplemental table S15). These effects often occurred at the onset of time restricted eating and tended to disappear as participants adapted.^{21 36}

Heterogeneity, inconsistency, and meta-regression

Heterogeneity and the corresponding inconsistency was generally not substantial (I²<75% for 87% of the investigated associations), except for HbA_1c and HOMA-IR (both I²>90%, representing 13% of the studied associations) (online supplemental tables S16 and S17). For HbA_1c, inconsistency was pronounced for all of the time restricted eating subtypes (I²=95.5%, τ²=0.11) and for the time restricted eating duration groups (I²=94.5%, τ²=0.10). Similarly, for HOMA-IR, inconsistency was high for the early, mid-, late, and self-selected time restricted eating groups (I²=99.3%, τ²=5.51) as well as for the eating duration groups (I²=99.4%, τ²=5.80). In meta-regression analyses, significant interactions (P<0.05) between time restricted eating subtypes and age, and between time restricted eating durations and body mass index, age, and intervention duration were observed for HOMA-IR but not for HbA_1c. In the late time restricted eating group, the treatment effect on HOMA-IR decreased with age compared with usual diets. For time restricted eating durations of <8 hours and eight hours, treatment effects on HOMA-IR decreased with age, whereas in the time restricted eating eight hour group, the effects increased with body mass index and intervention duration. We found no significant interactions for energy restriction and disease status. These results should be interpreted with caution because of the low number of studies for each treatment comparison.

Sensitivity analyses

In analyses restricted to healthy populations and intervention durations of 1-6 months, early time restricted eating consistently showed higher P score rankings for anthropometric measurements (except for fat free mass-lean mass) and glycaemic parameters, similar to the main analyses (online supplemental figures S16 and S17A). P score rankings for the different eating durations were mixed, similar to the main results (online supplemental figures S16 and S17B).

Discussion

Principal findings

Overall, time restricted eating was associated with significant improvements in a wide range of metabolic outcomes, including reductions in body weight, body mass index, fat mass, waist circumference, systolic blood pressure, fasting blood glucose levels, fasting insulin levels, and triglyceride concentrations, compared with usual diets. In the evaluation of P score ranking for various subtypes of time restricted eating, we consistently found that early time restricted eating and mid-time restricted eating showed higher rankings for anthropometric measurements (excluding fat free mass-lean mass) and glycaemic parameters. Compared with late time restricted eating, early time restricted eating significantly reduced body weight, waist circumference and fasting insulin levels, whereas mid-time restricted eating significantly decreased body weight, body mass index, and waist circumference. The relations between the duration of the eating window and metabolic outcomes were less consistent.

Effects of overall time restricted eating on metabolic health

A meta-analysis showed that compared with usual diets, time restricted eating had significant benefits for body composition, blood pressure, glucose control, and triglyceride levels, which aligns with our findings.⁶⁹ During time restricted eating, prolonged fasting leads to gradual depletion of glycogen stores. This shift in energy availability causes the body to transition from lipid and cholesterol synthesis and fat storage to mobilising fat through fatty acid oxidation and ketone production.⁷⁰ The metabolic improvements seen with time restricted eating can be partially attributed to reduced caloric intake and subsequent weight loss.^{71 72} Several studies, however, have shown that even without caloric restriction or notable weight loss, time restricted eating still results in favourable metabolic outcomes.^{17 73} These findings suggest that the metabolic benefits of time restricted eating are not only dependent on energy deficit or weight reduction, but that this unique eating pattern may exert independent physiological effects.

Effects of subtypes of time restricted eating on metabolic health

A meta-analysis compared early time restricted eating and non-early time restricted eating (including usual diet and late time restricted eating). The results showed that early time restricted eating significantly reduced fasting blood glucose levels and HOMA-IR, but had no significant effect on levels of triglycerides, total cholesterol, low density lipoprotein cholesterol, or high density lipoprotein cholesterol,⁷⁴ consistent with our findings. In another network meta-analysis comparing early time restricted eating, late time restricted eating, and non-time restricted eating, the analysis based on surface under the cumulative ranking (SUCRA) showed that early time restricted eating ranked the highest for body weight, fasting blood glucose levels, HOMA-IR, systolic blood pressure, and diastolic blood pressure, consistent with our results, except for diastolic blood pressure.⁷⁵ By including more literature and refining the classification of dietary patterns, our study suggests that early time restricted eating was superior in regulating blood sugar levels and reducing body weight, but the outcomes for blood lipid profiles were inconclusive.

The enhanced effectiveness associated with early time restricted eating in regulating blood sugar levels can be attributed to two potential factors: glucose tolerance, typically peaking in the morning and declining in the afternoon and evening,^{76 77} and insulin secretion rates, which peak from midday to late afternoon (1200-1800) and are lowest during sleep.⁷⁸ Therefore, consuming meals earlier in the day may promote synchronisation with eating behaviour and circadian rhythms, and hence promote metabolic health.⁷⁹ Jamshed et al suggested that early time restricted eating may improve insulin signalling in the morning through the enzyme phosphatidylinositol-3 kinase. This activation subsequently facilitates the translocation of glucose transporter type 4 to the plasma membrane, ultimately resulting in increased glucose uptake into the cell.⁸⁰ These activities could explain the favourable results for glycaemic parameters found in the early time restricted eating group in our analysis. Also, the greater reduction in body weight observed in the early time restricted eating group may be because of its potential to induce higher fatty acid oxidation in both adipose and muscle tissues.⁸¹

We found that self-selected time restricted eating showed favourable P score rankings for some outcomes, such as diastolic blood pressure, fasting blood glucose levels, and levels of triglycerides, total cholesterol, and low density lipoprotein cholesterol. Moreover, self-selected time restricted eating was significantly better than usual diets for body weight, fat mass, systolic blood pressure, and triglyceride levels. These beneficial effects, combined with the flexibility of self-selected time restricted eating, allowing individuals to select their own eating windows, indicate that this subtype of time restricted eating, although not superior to early time restricted eating overall, may also be a good strategy to promote adherence and improve metabolic health.

Most time restricted eating types significantly reduced fat free mass-lean mass compared with usual diets but no significant difference was seen compared with continuous energy restriction. Decreases in body weight are accompanied by a reduction in both fat mass and fat free mass-lean mass,⁸² and our results showed a larger decrease in fat mass than fat free mass-lean mass compared with usual diets. Future research should investigate whether time restricted eating can be further optimised (eg, by incorporating resistance training) to preserve fat free mass-lean mass.

Effects of time restricted eating duration on metabolic health

A meta-analysis compared different time restricted eating durations with non-time restricted eating. The findings showed that a shorter eating window (4-6 hours) was more effective in reducing body weight and fat mass but was not conducive to maintaining fat free mass. On the other hand, a moderate eating window (7-9 hours) seemed more advantageous for reducing fasting blood glucose levels, whereas a longer eating window (10-12 hours) seemed more beneficial for lowering levels of total cholesterol and triglycerides.⁸³ These findings are similar to our results.

In another network meta-analysis comparing different eating durations with non-time restricted eating, the results indicated that, based on SUCRA analysis, an eating window of ≤6 hours ranked higher in reducing body weight, fasting insulin levels, fasting blood glucose levels, and levels of triglycerides and low density lipoprotein cholesterol.⁸⁴ Our results, however, showed that a time restricted eating duration of <8 hours had a worse ranking than ≥8 hours in terms of fasting blood glucose levels, triglyceride levels, and low density lipoprotein cholesterol levels. Our findings might have higher certainty of evidence because we refined the classification of the control group to include different dietary strategies (eg, usual diet and continuous energy restriction), aiming to reduce heterogeneity within that group. Because of limited studies with extremely short or long eating durations, we categorised all studies with eating durations <8 hours into one group and durations >8 hours in another group. This approach avoided instances where some groups were based on only one study, which could potentially compromise the overall network transitivity assumption.

Effects of subtypes of time restricted eating and eating duration on metabolic health

Because of the interaction between the timing of a meal (early, mid-, or late time restricted eating) and circadian rhythms, as well as the effect of the duration of the eating-fasting cycle on physiological metabolic responses, both factors can collectively affect metabolic health. Therefore, we further investigated the combined effects of these two factors.

In our comprehensive analysis, although a time restricted eating duration of <8 hours showed the lowest P value rankings for lipid profile, a further breakdown showed that only early and mid-time restricted eating with a duration of ≤8 hours showed worse rankings, whereas this finding was not seen for late time restricted eating with a duration of ≤8 hours for levels of total cholesterol, low density lipoprotein cholesterol, and high density lipoprotein cholesterol. This discrepancy might be because most studies conducted blood analyses in the morning, resulting in longer fasting periods for early and mid-time restricted eating durations of ≤8 hours compared with late time restricted eating duration of ≤8 hours. This longer fasting period might prompt the liver to re-esterify free fatty acids released from fat cells, releasing very low density lipoprotein in response.⁸⁵

Another study suggested that early time restricted eating increased levels of total cholesterol, low density lipoprotein cholesterol, and high density lipoprotein cholesterol in the morning compared with a control diet, but this effect was not seen in the evening.⁸⁰ Therefore, in future studies, conducting a more detailed assessment of lipid profiles at different times and exploring lipid subfractions (eg, very low density lipoprotein and various molecular weights of high density lipoprotein cholesterol) will be crucial to confirm whether the changes in lipid profile have a pathophysiological basis.

Late time restricted eating with a duration of >8 hours generally showed the lowest P score rankings for most outcomes, but late time restricted eating with a duration of ≤8 hours showed a different pattern. For outcomes such as systolic blood pressure, diastolic blood pressure, fasting insulin levels, HOMA-IR, and levels of total cholesterol, low density lipoprotein cholesterol, and high density lipoprotein cholesterol, late time restricted eating with a duration of ≤8 hour ranked first or second. These findings suggest that both the timing of time and eating duration should be considered simultaneously during analysis.

Study implications

As well as discovering the benefits of early time restricted eating for metabolic health, we have also identified other unique features of time restricted eating and their significant clinical implications. Firstly, individuals showed higher compliance with time restricted eating, exceeding continuous energy restriction in most studies.^{19 32 42 48 58 63} Even without restricting start and end times, self-selected time restricted eating proved beneficial for metabolic health compared with usual diets. This finding underscores the flexibility of time restricted eating, allowing individuals to adjust to their personal circumstances, and hence enhance long term adherence and metabolic health.

Secondly, our findings indicated that time restricted eating has minimal associated side effects, which may diminish as participants become more accustomed to this dietary regimen,^{21 36} further emphasising the safety of implementing this dietary approach. Also, irrespective of the specific type of time restricted eating, systolic blood pressure was significantly decreased by 4-6 mm Hg compared with usual diets, which could translate to reduced mortality rates for ischaemic heart disease and cardiovascular disease of 8.6-13.2%, and a decrease in overall mortality of 5.8-8.6%.⁸⁶ Subsequent investigations of extended interventions and follow-up durations, and combinations with exercise, along with comprehensive ambulatory measurements or clinical outcomes, will give additional insights into the potential of time restricted eating as a dietary intervention to enhance both metabolic and overall health.

Strengths and limitations of this study

Our systematic review and network meta-analysis had several strengths. We reclassified the types of time restricted eating based on meal timing and eating duration to more precisely categorise different time restricted eating interventions across studies. We used network meta-analysis to combine direct and indirect evidence, allowing us to determine the most effective time restricted eating regimens for metabolic health. We explored the combined effects of timing of eating and eating duration by simultaneous assessment in a network meta-analysis, enabling us to more comprehensively understand the potentially differential effect of various time restricted eating regimens on metabolic health.

Our analysis was limited by the number of studies in this simultaneous analysis, particularly for time restricted eating regimens with limited direct evidence, such as late time restricted eating with an eating duration of <8 hours. Also, although we aimed to include outcomes such as metabolic syndrome and type 2 diabetes, no studies have yet investigated these outcomes. For the classification of time restricted eating durations (<8 hours, 8 hours, and >8 hours), although the eight hour window is the most commonly used protocol in the current literature,⁸⁷ we acknowledge that this categorisation has a degree of arbitrariness and the results should be interpreted with caution. Furthermore, although dietary interventions typically involve a lower risk of reporting bias than drug trials, potential bias may still exist. Therefore, our CINeMA certainty ratings may be optimistic and should be interpreted carefully. Although a network meta-analysis allows for the integration of direct and indirect evidence, more complex assumptions are involved. The scarcity of direct evidence is also a limitation, and further studies that directly compare the differential effects of time restricted eating subtypes are needed for more definitive conclusions. Therefore, the findings from our study should be interpreted with these assumptions and limitations in mind. Future research should look at these gaps to provide clearer clinical implications.

Conclusions

Overall, time restricted eating was associated with consistent improvements in body weight, body mass index, fat mass, waist circumference, systolic blood pressure, and levels of fasting blood glucose, fasting insulin, and triglycerides compared with usual diets. Early time restricted eating was superior to late time restricted eating, particularly for anthropometric and glycaemic parameters, but the association between duration of the eating window and metabolic outcomes was inconsistent.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.