What fasting does when your weight does not change
Almost every fasting study confounds two things: people fast, they lose weight, and their health markers improve. Which of those did the improving? One trial was designed to separate them, by the simple and expensive method of not letting anyone lose weight.
The design
Men with prediabetes were fed enough food to keep their weight steady, and randomised to either a 6-hour eating window with dinner before 15:00, or a 12-hour window. Five weeks each, then everyone crossed over to the other schedule. Every meal was supervised. Nobody lost weight, because nobody was allowed to.
Insulin sensitivity, β cell responsiveness, blood pressure and oxidative stress all improved on the early window — with weight held constant. That is the cleanest evidence available that meal timing does something in its own right, not only as a way of eating less.
How much this is worth
Enough to be interesting; not enough to redesign your life around.
It is one small proof-of-concept trial, in men, with prediabetes, over five weeks, on a window most people would find impossible to keep — dinner finished before three in the afternoon. A review covering the whole field lands somewhere more modest: across alternate-day fasting, the 5:2 pattern and time-restricted eating, the consistent findings are 1–8% weight loss and 10–30% reductions in energy intake, with blood pressure, insulin resistance and oxidative stress moving in the right direction. Cholesterol and triglyceride results were variable. The same review reports that intermittent fasting is generally safe and did not produce energy-level problems or increased disordered eating in the trials reviewed.
So: real effects, modest sizes, and mostly running through eating less.
Why this matters if the scale is stuck
Two honest reasons to keep a window even when your weight is not moving.
Some of the benefit is not weight. If your reason for fasting is blood pressure or blood sugar rather than a number on a scale, the controlled trial says the timing itself contributed something. A stalled scale does not mean nothing is happening.
A stalled scale is often not stalled. Daily weight is noisy enough to hide a real trend for weeks. Before concluding that a window has stopped working, read the trend rather than the reading — two 7-day averages a fortnight apart is the minimum honest comparison.
What this does not license
The trial above changed nothing about medication, and neither should you on the strength of it. Prediabetes and diabetes are managed with a doctor. The finding is that timing can move a marker; the finding is not that timing replaces treatment.
Related: fasting does not raise your metabolism and who should not fast.