Against ordinary dietary advice, twenty two trials of intermittent fasting moved weight by a third of one percent

What what a Cochrane review does differently is

A Cochrane review publishes its protocol before the results are known, has two independent reviewers apply a risk of bias tool to every trial, and attaches a certainty rating to every pooled estimate. When the reviewers rate their own finding low certainty, that is a verdict on the trials underneath it, not on the arithmetic.

Drawn from standard methodology, not from this paper.

Why they ran it

The reviewers set out to evaluate intermittent fasting against three different things at once: regular dietary advice, no intervention, and a waiting list. They note in their own background that fasting has been publicised in blogs and news articles while studies show inconsistent effects, and that this leaves physicians and patients uncertain.

Drawn from the review's stated objective and background.

Twenty two randomised and cluster randomised trials, 1,995 participants, all conducted in outpatient settings in North America, Australia, China, Denmark, Germany, Norway and Brazil, published between 2016 and 2024. Searches ran to 5 November 2024. Time restricted feeding, periodic fasting, alternate day fasting and modified alternate day fasting were all included. Minimum four week intervention and minimum six month follow up.

Against regular dietary advice, the difference in percentage weight loss from baseline was minus 0.33, with a confidence interval running from minus 0.92 to plus 0.26. Twenty one studies, 1,430 participants, low certainty.

Against no intervention or a waiting list, the difference was minus 3.42 percent, interval minus 4.95 to minus 1.90. Six studies, 427 participants, and this is the only moderate certainty estimate in the review.

Achieving a five percent reduction in body weight: risk ratio 0.98, interval 0.82 to 1.18, very low certainty. Quality of life: standardised mean difference 0.11, interval minus 0.27 to 0.49, low certainty.

Adverse events against dietary advice: risk ratio 1.45, interval 0.64 to 3.28, very low certainty.

None of the twenty two trials reported participant satisfaction, diabetes status, or any overall measure of comorbidity. All the evidence covers twelve months or less.

The numbers

Weight loss against regular dietary advice	minus 0.33 percent, 95 percent CI minus 0.92 to plus 0.26, low certainty
Weight loss against no intervention or waiting list	minus 3.42 percent, 95 percent CI minus 4.95 to minus 1.90, moderate certainty
Achieving five percent weight reduction	risk ratio 0.98, 95 percent CI 0.82 to 1.18, very low certainty
Quality of life	standardised mean difference 0.11, 95 percent CI minus 0.27 to 0.49, low certainty
Adverse events against dietary advice	risk ratio 1.45, 95 percent CI 0.64 to 3.28, very low certainty
Trials, participants, search cutoff	22 trials, 1,995 participants, searches to 5 November 2024
Outcomes never reported by any included trial	participant satisfaction, diabetes status, overall comorbidity

Why this might happen

Proposed by the authors This is the explanation the authors offer in their discussion. This study did not test it.

The reviewers state the mechanism they expect at the top of the review, that weight loss under fasting runs through caloric restriction plus increased fat metabolism and improved insulin sensitivity, and then find no weight difference against ordinary advice. Their own closing reading is not physiological at all. They tell physicians and patients to evaluate willingness and readiness, based on individual practicality and sustainability, which locates the remaining variable in whether a person can keep doing it.

Drawn from the review's background section and its stated conclusions.

What this does not show

It does not show that intermittent fasting fails to produce weight loss. It shows that the weight loss is the same size as the weight loss from ordinary dietary advice, and that both beat doing nothing. Those are different claims and the review is careful about the difference.
It does not settle whether fasting is safe. The adverse event interval runs from 0.64 to 3.28, which is compatible with fasting roughly halving harm or roughly tripling it. The reviewers rate that estimate very low certainty and so should anyone quoting it.
It says nothing beyond twelve months. The reviewers state this as a limitation in their own conclusions and ask for follow up periods extended past a year. Every included trial stops at or before twelve months.
It cannot tell you whether fasting helps someone with diabetes. Not one of the twenty two trials reported diabetes status as an outcome, which the reviewers flag as a gap in the literature rather than a finding.

Where this leaves us

The best conducted synthesis in this field finds that against ordinary dietary advice, intermittent fasting does approximately nothing, and against no advice at all it does what any diet does.

Adults with overweight or obesity in outpatient care across seven countries, in trials published between 2016 and 2024, none followed past twelve months.

Trials running past twelve months, and trials that report diabetes status and participant satisfaction, which no trial in this review did.

Caveats worth holding

Every certainty rating in the comparison against dietary advice is low or very low, and the reasons given are risk of bias, inconsistency and imprecision.
The moderate certainty estimate is the one against no intervention, which is the comparison least like a real clinical choice.
Trials were conducted in seven countries, all in outpatient settings, and the reviewers ask for future work in low and middle income countries and stratified by sex and body mass index category.