Randomised controlled trialindependent cross-checkfull text checked I-INS-09 in-depth review

Low-fat vs low-carb in a metabolic ward

Calorie for Calorie, Dietary Fat Restriction Results in More Body Fat Loss than Carbohydrate Restriction in People with Obesity.

Overview

Hall KD et al.. Cell Metab 2015 · PMID 26278052 · DOI 10.1016/j.cmet.2015.07.021

Cutting fat led to more fat loss than cutting carbs, even though cutting carbs lowered insulin more.

Does the study support the specific claim? No · Methodological quality: high

Original publication

Plain-language summary

Study explained simply

What did the researchers want to find out?: Does carbohydrate restriction (lower insulin) cause more fat loss than fat restriction at equal calories?
How was the study done?: Randomised crossover metabolic-ward study
How many people, animals or samples were studied?: 19 adults with obesity randomised (17 completed both diets)
What was actually found?: Fat loss 89 g/day on reduced-fat vs 53 g/day on reduced-carbohydrate (463 g vs 245 g over 6 days), although only the low-carb diet lowered insulin (by 22%).
How large was the observed effect?: +36 g/day with fat restriction
What do the results mean in everyday life?: Calories matter; insulin is not the only driver.
What are the study's limitations?: Only 6 days per diet; small sample; fat loss was a secondary outcome; long-term predictions come from a computer model; moderate (not very-low) carbohydrate restriction.
What does it say about Elawady's specific claim?: Contradicts C-INS-01 ('insulin the only cause of fat').
Do the results apply to Germany and Europe?: Physiology transferable.
What can explicitly not be concluded from this study?: Long-term weight outcomes.

Scientific in-depth review

1. Full reference

Authors: Hall KD et al.
Journal: Cell Metab
Year: 2015
Identifier: PMID 26278052 DOI 10.1016/j.cmet.2015.07.021
Verification: verified – PubMed efetch 2026-10-25 (this agent)
Retracted: no
Corrections: none
Extent checked: full text
Conflicts of interest and funding: Supported by the Intramural Research Program of the NIH (NIDDK). No competing interests statement in the retrieved text.

2. Research question

Does carbohydrate restriction (lower insulin) cause more fat loss than fat restriction at equal calories?

3. Study design and population

Study design: Randomised controlled trial – Randomised-order crossover controlled-feeding study on an inpatient metabolic ward (NCT00846040); two 2-week inpatient stays separated by a 2-4 week washout
Population: Adults with obesity (BMI >30; mean 35.9 +/- 1.1 kg/m2), 10 men and 9 women, weight-stable, non-diabetic per exclusion of interfering disease/medication
Country: USA (NIH)
Period: 2 x 6-day diets
Sample size: 19 randomised (10 M, 9 F); 17 completed both diet periods (2 men dropped out after the first stay on the RC diet)
Inclusion and exclusion criteria: BMI >30 kg/m2. Excluded: weight change >5 kg in past 6 months, menopause, pregnancy or breastfeeding, impaired mobility, diseases or medications interfering with outcomes, food or local-anaesthetic allergy, regular caffeine or alcohol use, eating or psychiatric disorders, vegetarian/kosher diets.

4. Methods and endpoints

Exposure / intervention: Reduced-carbohydrate (RC) diet: 30% energy reduction by 60% cut in carbohydrate (to ~29% of energy, ~140 g/d), fat and protein unchanged, for 6 days after a 5-day eucaloric baseline (50% CHO/35% fat/15% protein)
Comparator: Reduced-fat (RF) diet: isocaloric 30% energy reduction by 85% cut in fat (to ~8% of energy), carbohydrate and sugar unchanged
Endpoints: Primary: 24-h respiratory quotient (metabolic chamber). Secondary: energy expenditure, net fat/carbohydrate/protein oxidation, fat balance (fat intake minus net fat oxidation), cumulative body fat loss, DXA fat mass, body weight, 24-h urinary C-peptide (insulin secretion), fasting hormones and lipids
Statistical methods: Repeated-measures mixed model (compound symmetry) adjusted for sex and diet order; two-sided t-tests for time points; outliers by Cook's distance; no correction for multiple comparisons; retrospective minimal detectable differences (fat balance 220 kcal/d = 23 g/d; cumulative fat loss 110 g; DXA ~0.4 kg). Mathematical model used to design/power the study and simulate 6 months.

5. Actual results

Results: Only RC lowered 24-h insulin secretion (-22.3 +/- 7.0%, p=0.001) and increased net fat oxidation (+463 +/- 63 kcal/d). Rate of body fat loss: RF 89 +/- 6 g/d (840 +/- 60 kcal/d) vs RC 53 +/- 6 g/d (500 +/- 60 kcal/d), p=0.0002 in the results text (p=0.002 in the summary). Cumulative 6-day fat loss: RF 463 +/- 37 g vs RC 245 +/- 21 g (p<0.0001), ~80% greater with RF. DXA could not detect a between-diet difference in fat mass. Body weight loss was greater with RC (p=0.02), attributed to water. Model simulation: RF ~3 kg more fat loss after 6 months of perfect adherence; over a wide range of carbohydrate/fat ratios differences in fat loss are predicted to be small. Results reported as mean +/- SE; no 95% CIs given in text.
Absolute risks: not applicable (physiological outcomes; no clinical events)
Authors' conclusion: Calorie for calorie, fat restriction produced more body-fat loss; models predict small differences long-term.

6. Strengths

Inpatient metabolic ward with full control of food intake and weighed-back leftovers
Crossover design with randomised order and identical baseline diet before each period
Selective restriction of a single macronutrient with protein held constant; sugar not reduced on RF diet
Objective fat balance by 24-h indirect calorimetry, more sensitive than DXA
Pre-registered (NCT00846040) and model predictions published in advance

7. Weaknesses

Very short (6-day) diet periods; long-term inference rests on model simulation
Small sample (19 randomised, 17 complete crossovers)
Fat balance was a secondary outcome (primary was 24-h RQ); multiple outcomes without correction
Method of randomisation and allocation concealment not described
RC diet was moderate (~140 g/d), not very-low-carbohydrate or ketogenic
Highly artificial conditions (daily fixed exercise, chamber confinement); adherence-related real-world effects not addressed
Two protocol deviations (wrong diet on one day for two participants), retained in analysis

8. Risk of bias

Orienting tool: RoB 2 (crossover variant)
Type of assessment: structured assessment based on the full text by a single reviewer (no formal double review)
Full formal assessment: no
Overall risk of bias: some concerns
Rationale: Changed from low (abstract-based) to some_concerns because randomisation details are missing, headline outcome was secondary and there were 2 dropouts. The core finding is nonetheless robust. Single-reviewer structured judgement; a formal RoB 2 needs two independent reviewers and the protocol.

8. Risk of bias
Domain	Judgement	Rationale
D1 Randomisation process (incl. period/carryover)	some concerns	Order of RC and RF was randomised, but sequence generation and concealment are not described. Carryover unlikely given 2-4 week washout and repeated 5-day baseline diet; order included in the model.
D2 Deviations from intended interventions	low	Participants confined to ward, all food provided and weighed back; two single-day diet errors retained (removal did not change significance). Blinding not possible but adherence enforced.
D3 Missing outcome data	some concerns	2 of 19 (10.5%) dropped out after the RC period and contributed no RF data; mixed model used; reasons not stated.
D4 Measurement of the outcome	low	Fat balance and energy expenditure measured objectively by chamber calorimetry and urinary nitrogen.
D5 Selection of the reported result	some concerns	Registered with RQ as primary endpoint; fat balance (the headline result) was secondary; many outcomes without multiplicity correction, though fat-loss result is large and consistent with pre-published model predictions.

Full-text review (25 October 2026): Added inclusion criteria, 2 dropouts (17 completers), primary endpoint was RQ (fat loss secondary), insulin secretion -22.3%, cumulative fat loss 463 vs 245 g, p=0.0002 in text vs 0.002 in summary, DXA insensitivity, greater weight loss with RC, model predictions of small long-term differences; RoB 2 itemised, overall low -> some_concerns.

9. Assessment of the statistical interpretation

Within-subject difference of ~36 g/d fat loss exceeds the minimal detectable difference (23 g/d) and is unlikely to be chance, but it is a short-term physiological effect. The key mechanistic finding is that lowering insulin secretion by ~22% did not increase fat loss relative to an isocaloric diet that left insulin unchanged; the authors themselves predict only small long-term differences between isocaloric diets.

10. Relevance to the specific claim

Contradicts C-INS-01 ('insulin the only cause of fat').

11. Independent contrary findings

Carbohydrate-insulin model proponents.

12. Transferability to Germany and Europe

Physiology transferable.

13. Overall methodological judgement

High-quality, tightly controlled physiological experiment that directly falsifies the idea that lowering insulin is necessary for, or the only route to, body fat loss and shows calories matter; it does not establish which diet is better long-term in free-living people.

Multidimensional assessment
Multidimensional assessment
Authenticity of the source	verified
Methodological quality	high
Risk of bias	some concerns
Precision of results	moderate
Directness	indirect
Reproducibility	partly replicated
Consistency with the overall evidence	mixed
Regional transferability	high
Adequacy of the interpretation (by the archive)	not applicable

14. Open questions

Whether differences persist over months; very-low-carbohydrate/ketogenic comparisons (later studied by the same group); adherence in free-living settings; effects in people with diabetes.