Open questions (what we still don’t know)

Years, not weeks

In 2025, Zhila Semnani-Azad, Frank Hu and colleagues reviewed 99 randomized trials (6,582 adults) of intermittent fasting, ordinary calorie cutting, and unrestricted eating. The outcomes were body weight and short-term numbers such as glucose, insulin, and blood lipids. Most of those trials lasted less than 24 weeks. Seventeen lasted 24 weeks or more. That set was not built to count heart attacks, strokes, or deaths over many years.

A survey can follow deaths for years and still be a different kind of study. In 2025, Meng Chen, Lan Xu, Victor W. Zhong and colleagues took two 24-hour diet recalls from 19,831 adults in the U.S. National Health and Nutrition Examination Survey, from 2003 to 2018. They kept the people who said those days were their usual intake, and linked the recalls to deaths through December 2019. Over a median of 8.1 years, an eating window shorter than 8 hours went with a higher rate of death from cardiovascular disease than an eating window of 12 to 14 hours. Their estimate was 2.35 times that rate, and the published range around it ran from 1.39 to 3.98, which is wide. Death from any cause, and death from cancer, did not show the same link. The authors say the study cannot tell whether the short window caused the deaths. People who already eat in a very short window can also be sicker, smoke more, sleep less, work different hours, or have less money. The higher rate of cardiovascular death was still there when the authors split the results by income and similar circumstances, and by smoking. Two diet recalls are not a trial that assigned a 16:8 window and followed it for years. A 16:8 day is 8 hours of eating. The higher rate in that paper sits with windows shorter than 8 hours.

There is still no randomized trial that assigns a daily window, keeps it for years, and counts heart attacks, strokes, and death from any cause against ordinary eating or against eating less without a clock. Work in animals on lifespan answers a different question. People sometimes bring a longevity hope to the clock. These trials have not measured that hope.

Who was never enrolled

The people in these randomized trials are mostly adults with extra weight. Pregnancy, age under 18, underweight, a current eating-disorder diagnosis, and insulin-treated diabetes are often left out. The results describe the adults who enrolled.

Older adults who cannot afford to lose lean mass, people who already eat little, shift workers, and anyone whose medicines need food are still rarely enrolled. So are children and adolescents. The checklist for those groups is Who should be careful with fasting. A missing study is not a clearance to try the hours on yourself.

These trials do not supply a menstrual-cycle plan, a polycystic ovary syndrome (PCOS) plan, or a testosterone plan. Reports people give, and the point at which a clinician comes first, are in Fasting, periods, and hormones.

The clock and eating less

Compared with unrestricted eating, a shorter window often means people eat less, and weight drops by a few kilograms over weeks to months. Compared with eating less on purpose and no special clock, Semnani-Azad and Hu found the extra effect of the hours was small, especially in the trials that lasted 24 weeks or more. Alternate-day fasting was the only intermittent style with a small extra drop against continuous calorie cutting in the shorter trials, about 1.3 kilograms, and that extra mostly faded in the longer ones.

What remains open is how much of a successful window was simply eating less, in people living at home for years, and how much was the clock itself. Trials that hold calories steady are fewer and shorter than the weight trials. In 2018, Elizabeth Sutton, Courtney Peterson and colleagues fed eight men with prediabetes every meal under supervision. Each man spent five weeks on an early six-hour schedule, with dinner before 3 p.m., and five weeks on a twelve-hour schedule, with enough food to hold weight steady. Insulin sensitivity improved on the early short schedule, and blood pressure was lower. Eight supervised men is a real finding and a narrow one. It is not a 16:8 window kept at home for years.

What people eat inside the window is still measured less often than the clock. A set of hours does not repair a very large meal or a plate with too little protein. Trials that only set when people eat cannot rank 16:8 against eating less of the same foods.

When the eating hours sit

In 2026, Yu-En Chen, Hui-Li Tsai and colleagues pooled 41 randomized trials (2,287 people) of daily eating windows that lasted longer than a month. The search ran through the start of 2023, so the 2026 date is when the review was published. Early meant the last meal before 5 p.m. Late meant the last meal after 7 p.m. Compared with the late windows, the early windows had lower fasting insulin, by about 3.3 microunits per millilitre, and lower body weight, by about 1.2 kilograms. They rated both differences high certainty. The insulin gap and the weight gap came as a pair. How long the window lasted, under eight hours, about eight hours, or longer than eight hours, did not line up as better or worse in a consistent way.

A direct week of early hours versus later hours looks quieter. In 2019, Amy Hutchison, Leonie Heilbronn and colleagues had fifteen men at risk of type 2 diabetes eat from 8 a.m. to 5 p.m. for one week and from noon to 9 p.m. for another week. The glucose rise after a test meal was smaller in both weeks than in their usual-eating week. The morning window and the afternoon window were not different from each other. One week and fifteen men does not make a rule that dinner has to move.

Shift work, a household that eats together at 8 p.m., and a person who is hungriest in the morning sit outside those comparisons. A slightly lower lab number that wrecks sleep is a worse week. Restaurants, a night shift, and a late table are the subject of Keeping a window in real life.

Muscle, training, and many years

Lean mass can fall when someone eats less energy than they use, with a clock or without one. In 2026, Kaijun Xing, Yannan Zhang and colleagues pooled 28 randomized trials by age. Fat-free mass, muscle and other tissue that is not body fat, was down under both daily windows and intermittent energy restriction, on the order of about a kilogram. Certainty was stronger in some age groups than others. Not every time-restricted-eating subset shows that drop, and scanners are imperfect. That kilogram of fat-free mass is a different result from lost strength, and from lost body fat. It is not a decade of follow-up in older adults.

In 2025, dietitian Kelsey Gabel and colleagues reviewed 23 trials that combined intermittent fasting with exercise. Body fat usually went down. Muscle and strength were mixed, and the mix depended on who enrolled and how much they trained. Papers on lifting while hungry are fewer. In 2025, Alexandra Ferreira Vieira and colleagues found four such trials, and three had a high risk of bias. Four small trials leave both hungry lifting and lifting after a meal as open questions. Whether enough protein and some lifting spare muscle over many years, in older adults, is still a question with few studies. People who felt weak and left are often missing from the averages.

Autophagy and ketones in people

Autophagy is a real process: cells break down worn parts and reuse the pieces. It runs at a baseline all the time. In 2016, Yoshinori Ohsumi received the Nobel Prize in Physiology or Medicine for work on that machinery in yeast. The prize is about how the process works in cells. It is not a result for a 16-hour window in a person.

Researchers infer autophagy with tissue samples and protein markers in a lab. Those tools are imperfect even there. A stage name on a timer, including the name that sits around 24 to 48 hours, is an hour range. There is still no simple human test a clock could use to say that autophagy is underway.

Ketones rise as the carbohydrate available for fuel falls. That marks fuel during the fast. It is a different number from body fat at the end of the week, and it is not a health score. A fast of several days has fewer human studies, and those studies are more often run with a clinic nearby. Adding those hours does not answer whether a daily window changes deaths over many years.

Staying with the hours

Whether people keep the assigned hours is part of the result. People who quit leave the average, so the number that remains can look calmer than the week that made someone stop. In the Semnani-Azad and Hu review, the small extra differences versus ordinary calorie cutting mostly faded once trials reached 24 weeks or more. Longer trials look quieter because the eating hours drift after the research staff stop checking in.

What is still missing is which daily windows people are still keeping five years later, without the clock becoming a streak or a reason to skip the only shared meal. Shortening the window, or skipping a day, is part of keeping a pattern a person can still recognize. The papers have not followed that for five years.

What a missing answer still allows

A few kilograms, or a small move in a lab number, over weeks or months in a specific group of adults, is still a result for those weeks. A lifetime plan would need the years these trials did not run.

Ending dinner a little earlier can be a choice about sleep and the shape of the evening. That choice can stand on its own.

A timer with no time cap is room to stop when you need to. A 48-hour fast answers a different question from the one a 12-week daily window left open. None of these results is a diabetes treatment, a longevity plan, or a hormone plan, and none of them is a reason to add a scale.

The timer

The clock counts elapsed hours. Stage names are hour ranges. Custom, set to 12 or 14 hours, 16:8, or Open are labels for a plan you already chose. You can run the clock in the browser.

Warning signs still end the fast. They are written out in When to stop a fast. Medicines that need food, or that lower glucose, stay with the prescriber.

For this week, a window you can keep, and enough food in the hours you eat, still matters more than treating a missing study as instructions.