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

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). 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