Time-restricted eating lowered fasting glucose, not HbA1c

Time-restricted eating has become one of the most popular dietary strategies of the past decade, promoted on the promise that compressing when you eat improves how your body handles glucose. That promise matters most for people who already have impaired glucose control — prediabetes or type 2 diabetes — because for them a genuine improvement could change disease trajectory. But much of the enthusiasm rests on small trials with short durations and mixed endpoints, and glucose control can be measured in several ways that do not always agree. This review asked what the randomised evidence shows across the markers clinicians actually use.

The authors searched MEDLINE, Embase and CENTRAL from inception to 5 August 2025, plus forward and backward citation searches, screening 2043 database records and 1249 further citations. Eligible studies were randomised trials in adults with prediabetes or type 2 diabetes, lasting at least 2 weeks, comparing an eating window of 12 hours or less against a non-time-restricted control diet. Ten trials with 599 participants qualified; mean study length was 4 months and eating windows ranged from 4 to 10 hours per day. Data were pooled as weighted mean differences using random-effects models, with GRADE for certainty and Cochrane RoB 2 for risk of bias.

The pooled analysis showed no significant overall effect on HbA1c (-3.33 mmol/mol; 95% CI -6.87 to 0.20; p=0.06, moderate certainty). Stratified by group, HbA1c fell by 0.93 mmol/mol in people with prediabetes (p=0.02) but not in those with type 2 diabetes (-4.68 mmol/mol; p=0.09). Fasting blood glucose did fall overall by 0.30 mmol/l (p<0.01, moderate certainty), and in both subgroups: -0.14 mmol/l in prediabetes and -0.48 mmol/l in type 2 diabetes. Body weight dropped by 1.6 kg in the pooled analysis, while effects on insulin sensitivity, beta cell function and continuous glucose monitoring measures were inconclusive.

This tempers the strongest claims made for meal timing without dismissing the approach. Fasting glucose responded consistently and with moderate certainty, but HbA1c — the marker that reflects average glucose over months and drives clinical decisions — did not move overall, and the significant prediabetes reduction was small in absolute terms. The pattern raises the question of whether the modest weight loss explains much of the glucose effect, which these data cannot resolve. Imprecision from small samples, and inconclusive results for insulin sensitivity and beta cell function, leave the mechanism open; the authors call for large-scale trials of long-term effects in both prevention and treatment.

If you are thinking of compressing your eating window to improve blood sugar, this is a reason to expect a real but modest effect on fasting glucose and body weight rather than a transformation of your longer-term glucose control. The lack of an overall HbA1c change is the key honest detail, and the small reduction seen in prediabetes should be read for what it is: small. Nothing here shows that timing beats simply eating less or better, since weight also fell. Anyone managing prediabetes or type 2 diabetes should treat this as context for a discussion with their clinician rather than a reason to change medication or meal patterns unsupervised.