Caloric Restriction and Time-Restricted Eating: What the Trials Show

Caloric restriction is the most durable finding in the biology of aging. Eat less, live longer — demonstrated in yeast, worms, flies, rodents, and with more argument, primates.

Time-restricted eating arrived later and spread faster, because it offers the promise of the same benefits without the hard part: don’t eat less, just eat within a window.

We now have controlled human trials on both. They do not say the same thing.

CALERIE: what 12% restriction actually does

CALERIE was a phase 2 randomized trial in healthy, non-obese adults aged 21–50 with BMI 22.0–27.9. Participants were assigned 2:1 to a prescribed 25% calorie restriction or an ad libitum control diet, for two years. Cardiometabolic outcomes were evaluated in 218 participants (Kraus et al., 2019).

The first honest finding is about adherence. The prescription was 25%. The achieved restriction was 11.9% — roughly 280 kcal per day. Even in a supervised, multi-centre trial with substantial support, participants delivered less than half the intended deficit.

That produced a sustained 10% weight loss over two years, of which 71% was fat mass. Retention was good: 82% of the restriction group completed.

And the markers moved, essentially across the board:

• LDL cholesterol fell about 7% (2.51 to 2.33 mmol/L)
• Systolic and diastolic blood pressure both decreased, evident by 6 months and significant at 12
• C-reactive protein decreased, significant at year 2
• Fasting insulin fell by about 1.71 μIU/mL at 24 months; the insulin sensitivity index rose significantly
• HOMA-IR improved by about 0.36 at two years

That is a coherent, favourable shift in almost every risk factor that matters, produced by an achieved deficit of about 280 calories a day in people who were not overweight to begin with.

The intervention that worked was smaller than the one prescribed. That is the most useful practical detail in the trial.

What CALERIE does not tell you is whether any of this extends life. It was two years long in healthy adults. The outcomes are risk factors, not events. The authors also note they lacked clinical measurement of atherosclerotic plaque change.

TREAT: time-restricted eating, tested properly

TREAT randomized 116 adults with overweight or obesity to 16:8 time-restricted eating — all calories between noon and 8pm — or three structured meals per day, for 12 weeks. 105 completed (Lowe et al., 2020).

Weight change:

• Time-restricted eating: −0.94 kg (95% CI −1.68 to −0.20; P = .01)
• Control: −0.68 kg (95% CI −1.41 to 0.05; P = .07)
• Between-group difference: −0.26 kg, P = .63

Not significant. Under a quarter of a kilogram apart after twelve weeks.

Then the finding that generated the most discussion. Appendicular lean mass index fell by −0.22 kg/m² in the time-restricted group versus −0.06 kg/m² in controls, a significant between-group difference of −0.16 kg/m² (P = .005). Roughly 65% of the weight lost in the time-restricted arm was lean mass — far above the 20–30% typically expected during weight loss.

The authors concluded that time-restricted eating, in the absence of other interventions, is not more effective for weight loss than eating throughout the day, and did not improve metabolic markers.

The limitations deserve equal weight. The trial did not directly measure caloric or protein intake, relying on mathematical modelling. It did not assess whether protein intake changed, which is a substantial gap given the lean mass result. And DXA lean-mass analysis does not account for variation in muscle hydration. The lean mass finding is a serious signal, not a settled fact.

Twelve weeks is also short, and the population had overweight or obesity, so this does not test every version of the practice.

The variable both trials point at

Read together, the two trials suggest the operative variable is energy intake, not clock time. CALERIE restricted energy and moved everything. TREAT restricted timing without enforcing energy restriction and moved almost nothing.

Which raises the question of what actually determines how much people eat when nobody is counting.

A tightly controlled inpatient crossover trial gives the clearest answer available. Twenty weight-stable adults ate ad libitum on an ultra-processed diet for two weeks and an unprocessed diet for two weeks, in random order. The diets were matched for presented calories, sugar, fat, sodium, fibre and macronutrients (Hall et al., 2019).

Participants ate 508 ± 106 kcal per day more on the ultra-processed diet (P = 0.0001). They gained 0.9 kg on it and lost 0.9 kg on the unprocessed diet.

Same nutrients on paper. Five hundred calories a day of difference in what people spontaneously ate.

The sample was 20 people over four weeks. It is small. It is also one of the very few metabolic-ward randomized crossover trials in nutrition, which is the strongest design available for this question.

What to do with this

• Target energy intake, not eating windows. The trial that moved LDL, blood pressure, CRP, fasting insulin and HOMA-IR restricted calories. The trial that restricted timing alone moved a quarter of a kilogram.
• Expect to achieve about half of any deficit you prescribe yourself. CALERIE prescribed 25% and got 11.9% under close supervision. Plan for that gap instead of being defeated by it.
• Note that 12% was enough. A ~280 kcal daily deficit in non-obese adults produced broad cardiometabolic improvement over two years. The useful intervention is smaller and more sustainable than the heroic version.
• If you use a 16:8 window, protect protein and resistance training. The TREAT lean-mass signal is the one finding that should change behaviour: compressing your eating window makes it easier to under-eat protein and easier to lose muscle.
• Change food composition before changing timing. A 508 kcal/day spontaneous difference from processing alone, at matched macronutrients, is a larger lever than any window schedule tested so far.
• Do not extrapolate to lifespan. No human trial has shown that caloric restriction extends life. CALERIE showed it improves risk factors over two years. Those are different claims and the second does not guarantee the first.

Time-restricted eating is not useless. For some people it is a workable way to eat less, and eating less is the thing that worked. But it is a delivery mechanism for a calorie deficit, not an alternative to one — and the head-to-head trial is the reason to say so plainly.