Meal timing

The circadian system organises metabolism into 24-hour cycles, separating opposing processes in time and anticipating recurring periods of feeding and fasting. Human studies using a range of laboratory protocols report circadian rhythms in blood glucose, insulin, glucose tolerance, lipid levels, energy expenditure and appetite, with several peaking in the biological morning or around noon. Circadian misalignment, caused by light exposure, sleep or food intake at the wrong biological time, impairs metabolism in humans and has been linked to the development of type 2 diabetes, obesity and high blood lipids [1]. This provides a plausible biological basis for effects of meal timing that are independent of what or how much is eaten, although it does not by itself establish the size of any effect on disease.

The largest study of meal timing and death used data from 31,044 adults aged 40 and over in the US National Health and Nutrition Examination Survey (NHANES), collected between 1999 and 2018, with meal times estimated from a single 24-hour dietary recall. Over a median follow-up of 8.6 years, 7,129 participants died. Compared with a first meal between 7:00 and 8:00, all-cause mortality was 10% higher for a first meal between 8:00 and 10:00, 19% higher between 10:00 and 12:00, and 29% higher after 12:00. Compared with a last meal between 19:00 and 20:00, mortality was 13% higher for a last meal before 19:00 and 27% higher for eating after midnight. Similar associations were seen for cardiovascular mortality. At age 50, estimated life expectancy was 2.46 years shorter with a first meal after 12:00 and 2.35 years shorter with a last meal after midnight than in the reference groups [2].

The relationship for the last meal was therefore U-shaped rather than linear: both an unusually early and a very late last meal were associated with higher mortality than a mid-evening one [2].

The French NutriNet-Santé study followed 103,389 adults for a median of 7.2 years, with meal times taken from repeated 24-hour dietary records. A later first meal (after 9 a.m. compared with before 8 a.m.) and a later last meal (after 9 p.m. compared with before 8 p.m.) were associated with higher risk of cardiovascular outcomes, especially among women. For cerebrovascular disease, a last meal after 9 p.m. carried a hazard ratio (the risk in one group relative to another) of 1.28 (95% CI 1.05 to 1.55) compared with a last meal before 8 p.m., and each hour later carried a hazard ratio of 1.08. No association was found between meal timing and coronary heart disease, or between the number of eating occasions and cardiovascular risk. The authors concluded that a longer overnight fast combined with an early last meal, rather than skipping breakfast, may be useful for preventing cardiovascular disease [3].

The Japan Collaborative Cohort Study followed 28,625 men and 43,213 women aged 40 to 79, free of cardiovascular disease and cancer at baseline, for 19 years, during which 4,706 died of cardiovascular disease. Participants were grouped by supper before 8 p.m., after 8 p.m., or at irregular times. Compared with early supper, irregular supper timing was associated with 44% higher mortality from haemorrhagic stroke (bleeding into the brain). Supper timing showed no significant association with other types of stroke, coronary heart disease or total cardiovascular mortality [4]. Cardiovascular conditions are described under cardiovascular disease.

Time-restricted eating limits all food intake to a fixed window each day; early time-restricted eating places that window early in the day, in line with the circadian rhythms in metabolism. The first supervised controlled feeding trial of this approach randomised men with prediabetes to a 6-hour eating window with dinner before 3 p.m. or a 12-hour control window for five weeks, then switched them to the other schedule. Participants were fed enough food to keep their weight stable, so that any effects could not be attributed to weight loss [5].

Eight overweight men, aged 56 on average, completed the trial. The early schedule improved insulin sensitivity, the responsiveness of insulin-producing beta cells, oxidative stress and appetite, and lowered morning systolic and diastolic blood pressure by 11 mmHg and 10 mmHg relative to the control schedule. The authors described the trial as a proof-of-concept study and noted its small size as a main limitation [5]. A 6-hour window ending in mid-afternoon is considerably stricter than an ordinary early evening meal, so the trial tests a different practice from those measured in the observational studies. Blood sugar regulation is described under metabolic health and type 2 diabetes.

A randomised crossover trial (in which each participant completes every condition in random order) tested the effect of dinner timing on sleep in 13 healthy women with an average age of 21.4. Dinner of 709 to 740 kcal was eaten either 1 hour or 5 hours before habitual bedtime for four days in each condition, with sleep recorded by overnight electroencephalography (EEG, a recording of the brain's electrical activity). On the fourth day, the late dinner produced significantly shorter total sleep time, lower sleep efficiency (the share of time in bed spent asleep), more time awake after first falling asleep, a higher arousal index and more shifts between sleep stages. Post-meal glucose showed only transient differences, and a glucose tolerance test the next morning did not differ between conditions [6]. Sleep architecture is described under sleep regulation and sleep stages and sleep problems under insomnia.

The observational studies estimated meal times from self-reported 24-hour recalls or food records, in NHANES from a single day, which may not reflect usual habits [2]. They cannot exclude reverse causation (illness or other circumstances changing eating times, rather than eating times causing illness): shift work, sleep disorders, depression and undiagnosed disease can each shift or disrupt meal times and independently affect health. This may help explain why the earliest last-meal group in NHANES had higher mortality than the mid-evening group. The studies also used different cut-off times, 19:00 to 20:00 in NHANES, before 8 p.m. and after 9 p.m. in NutriNet-Santé, and before or after 8 p.m. in the Japanese study, so their estimates are not directly comparable [2, 3, 4]. The randomised trials were small and short, and none measured disease or death [5, 6].

Full deep dive: Meal timing and an early last meal.