FitChef

Does Volume Eating Work? What a 31-Study Meta-Analysis Found

31 studies tested what happens when you eat the same amount of food but lower its calorie density. The answer reframes everything you thought you knew about eating less.

Meta-analysisRobinson et al.International Journal of Behavioral Nutrition and Physical Activity202231 people7 min read

Drafted with AI from the full paper. Checked by a person.

Every portion control tool you have ever used runs on the same assumption: the only way to eat fewer calories is to eat less food.

Smaller plates. Measured servings. The bag that says 100 calories. And every time, the same result — you eat the correct amount, the hunger arrives, and an hour later you are back at the fridge wondering why you bother.

A research team led by Eric Robinson at the University of Liverpool pulled together 31 controlled feeding studies spanning three decades of nutrition research.

Across 1,086 participants — mostly adults, both sexes, healthy weight to obese — scientists had tested one question. What happens when you change the calorie density of the food on the plate without changing how much food is on it?

The result was large, consistent, and so far from chance that the probability of the pattern being noise was less than 1 in 1,000. Lowering calorie density reliably reduced how much people consumed in a day — and the reduction was not small.

But the question that no plate comparison on the internet has ever answered with data is not whether volume eating works. It is how much. And when Robinson's team isolated the studies where every meal was swapped to a lower-density version, the number that came back changed the scale of the conversation entirely.

Swap every meal to a lower-calorie-density version and you consume 709 fewer calories a day — not by eating less food, but by eating the same volume of food that carries fewer calories per bite. The math on that deficit: roughly 1.4 pounds a week. From a swap, not a sacrifice.

Robinson et al. 2022 · 31 studies, 1,086 participants

Key takeaways

Lower the calorie density of what you eat and the savings scale, your body barely pushes back, and the method you use to get there does not matter.

  1. The body claws back roughly 35 extra calories after a lower-density meal — a compensation so small it was not distinguishable from zero across 31 studies.
  2. The calorie savings scale linearly — change one meal and the effect is measurable, change every meal and it multiplies.
  3. Any approach to lowering calorie density works — adding vegetables, reducing fat, or using less calorie-dense versions of the same foods all produced comparable results.
  4. Portion control produces similar calorie reductions but triggers more compensation, making the savings harder to sustain.

What Changes When the Plate Stays Full

In eight of the 31 studies, researchers went all-in. Every meal participants ate was swapped to a lower-calorie-density version — same foods, same textures, comparable volume — prepared with more vegetables, less added fat, or ingredients that carry fewer calories per gram.

The average daily calorie savings across those eight studies: 709 calories.

That is roughly the calorie content of an entire meal. Gone from the day's total. Not because anyone ate less food — because the food they ate carried less energy per bite.

Run the math on a 709-calorie daily deficit and you land at roughly 1.4 pounds per week. That is a mathematical translation of the deficit, not a weight-loss promise — Robinson's team measured what people ate, not what they weighed over time.

But the scale of the intake reduction reveals something no portion plate ever could: the size of the opportunity sitting inside the same volume of food.

And the eight all-meals studies were the extreme case. The remaining studies tested single meals or partial swaps. Even partial changes produced measurable drops in daily intake. The pattern held: change the density, and intake follows.

But anyone who has tried any diet trick knows the next question before the researchers even ask it. Does your body fight back?

The Catch That Barely Exists

This is the question that sits between you and every diet promise you have ever heard. You eat less at lunch. Your body ramps up hunger. You eat more at dinner. The deficit disappears.

The catch is the reason every restriction trick eventually fails — your body compensates. And the assumption behind every warning about volume eating is that the same thing will happen here.

Robinson's team measured exactly that. Across the 31 studies, they tracked not just what happened at the manipulated meal but what happened at every eating occasion afterward.

Thirty-five calories.

That was the average compensation. Across 31 studies and more than a thousand participants, people ate 35 extra calories later in the day. A number so small it was not even distinguishable from zero in the analysis.

That means roughly 90 percent of the calorie savings persisted. Save 330 calories at a single lower-density meal, and your body claws back 35 of them later. The other 295 are gone.

Swap every meal to lower calorie density and daily intake drops by 709 calories — without eating less food. Not smaller portions. Not fewer meals. The same plates, rearranged.

Every friend who ever told you "your body will just eat more later" was describing a fear, not a finding. Across three decades of controlled research, the compensation everyone warned you about amounts to a rounding error.

One caveat the data cannot dismiss: these studies tested people in metabolic equilibrium, not people recovering from a diet. After weight loss, eight of nine appetite hormones remain altered for at least 62 weeks — a hormonal landscape none of Robinson's 31 trials were designed to measure. Whether the 90% savings figure holds when the hunger system is already in overdrive is a question the current evidence leaves open.

And that finding — that the body barely pushes back — is what makes the next comparison so devastating for the approach most people have been using.

THE CATCH
90%of savings kept
35cal
Robinson et al., 2022 · 31 studies · 1,086 participants

Why Smaller Plates Fight Back

If lowering calorie density works this well and the body barely fights back, Robinson's team faced the obvious next question. How does it compare to the approach most people actually use?

Portion control — eating smaller amounts of the same food — has its own body of research. Robinson's analysis noted that the effect sizes are comparable. Reducing portion size and reducing calorie density both produce meaningful drops in daily calorie intake.

But the similarity ends at the top-line result. In Robinson's own Discussion, the researchers cited evidence that portion control produces "more substantial compensation." When you shrink the plate, the body pushes back harder. When you change what is on the plate, it barely notices.

Same destination. Different resistance.

That difference is not abstract if you have ever sat across from a plate you knew was the right size and felt your stomach disagreeing before you finished. That disagreement has a name now — it is compensation.

And it hits harder when you shrink the plate than when you change what is on it.

The density principle works at the plate level — but inside that plate, one component has its own appetite-suppression pathway. Viscous fibers like beta-glucan and psyllium form gels that slow gastric emptying and trigger gut-stretch signals, mechanisms that operate independently of calorie density. Adding them to a lower-density meal may stack two suppression effects that use different biological channels.

But before the case for density over portions becomes too clean, Robinson's team was careful to show exactly where the evidence gets uncertain.

What the Lab Cannot Promise

Every one of the 31 studies took place in a controlled laboratory setting. Participants ate food selected, prepared, and measured by scientists. Study durations ranged from a single day to two weeks.

Robinson's team acknowledged this directly. In real kitchens, with real grocery stores and social meals, compensation would likely be greater — their own words, rooted in the recognition that lab conditions reduce the noise real life introduces.

The variation between studies tells a similar story. Even after removing statistical outliers, the spread across results was moderate. The direction of the effect was remarkably consistent — 29 of 31 studies found calorie intake went down when density dropped.

But the magnitude varied — some studies saw massive reductions, others modest ones.

The number 709 is a point estimate from the all-foods subset. It is not a guarantee of what happens in your kitchen on a Wednesday. Individual responses will differ, and the researchers had the integrity to say so.

And body weight itself? Robinson's analysis found a trend of roughly 1.5 pounds lost, but the result was not large enough to be definitive. These studies proved that density manipulation reduces what people eat. Whether that reliably translates into lasting body weight change was not something their short-duration data could answer.

This honesty is the reason the data is trustworthy. A page that hides the caveats is selling you something. This page is showing you what the research found — including the parts that remain uncertain.

A Dial, Not a Switch

The dose-response relationship across Robinson's data was linear. For every 100 calories of food served at lower calorie density, participants consumed roughly 77 fewer calories.

That linearity transforms volume eating from a headline into a tool. You do not need to overhaul every meal.

The data suggests that changing a single meal saves in the neighbourhood of 200 calories. Change two and it roughly doubles. Go all the way and you reach the territory of 709.

This is not an all-or-nothing deal. It is a dial you turn as far as you want, and the savings scale with you.

FOR EVERY 100 CALORIES CHANGED
77calories stayed saved
Robinson et al., 2022 · 31 studies · 1,086 participants

A Principle, Not a Plan

One more finding removes the last barrier. Robinson's team compared two approaches to lowering calorie density. The first: altering macronutrient composition — adding more vegetables, reducing fat. The second: keeping macronutrient ratios the same but using less calorie-dense versions of the same foods.

The difference between the two approaches? No meaningful difference. Both worked. The effect on daily calorie intake was comparable regardless of which path people took to lower density.

That means the remaining objection — "I do not want to eat nothing but salads" — does not survive the data. A lower-density version of the food you already enjoy produces the same effect as rebuilding the plate from scratch. The method does not matter. The principle does.

Robinson's analysis is not the only evidence pointing this direction. When Holt and colleagues tested 38 common foods for how full they left people after eating the same number of calories 1, the results confirmed the principle from a different angle.

The hunger after a measured portion was never a willpower failure. Robinson's data shows the body pushes back harder when you shrink portions than when you lower density. The resistance was in the approach — not in you.

Foods with the lowest calorie density consistently produced the highest fullness scores. Boiled potatoes scored seven times higher than croissants.

Thirty-one studies, three decades of controlled research, and converging evidence from independent satiety science all point to the same conclusion. The path to eating fewer calories was never about eating less food. It was about eating food that carries less energy per bite — and doing it any way that fits your life. Density was the single strongest lever — but the full guide found six more, and the interplay between them determines what actually holds.

That answers the density question. But it opens another. If the volume and density of your food control how much you consume at the plate level, is there a separate mechanism at the macronutrient level — one that explains why some nutrients suppress appetite more powerfully than others?

What it means for you

What this means · what other research found · your line · before you change anything

What this means

Robinson's data reframes the daily food decision. The question was never "how much should I eat?" It was always "what calorie density am I eating at?"

That reframe changes what matters at every meal. The plate stays full. The portion stays the same. The only thing that shifts is the composition.

What this means for you

Which one is you?

If you want to start with one meal

The all-meals figure grabbed the headline. But the single-meal data is where most people will start.

Studies that swapped just one meal found calorie savings of roughly 208 calories per day. Every additional swap added a proportional reduction. Starting with one meal is not a half-measure — it is the entry point on a curve that scales as far as you take it.

From what the study found.

If you are currently doing portion control

The hunger you feel forty minutes after a measured portion has a name in Robinson's data: compensation. And it hits harder when you shrink the plate than when you change what is on it.

Not a willpower problem. A strategy problem. The data suggests the path with less resistance was always available.

From what the study found.

Feeding a family

Four of the 31 studies tested children specifically. The effect direction held: children also ate fewer calories when food density dropped. The effect was slightly smaller than in adults, but the difference was not statistically meaningful.

The moderation analysis confirmed this — density changes at the family table are not just for the parent watching their weight.

The caveat: only four child studies. The direction is clear but the precision is limited. Robinson’s team flagged this as a gap needing more research.

From what the study found.

What other research found

confirmsHolt et al. 1995

41 participants

When 38 common foods were tested for how full they left people after eating the same number of calories, the foods with the lowest calorie density consistently produced the highest fullness scores — with protein, fiber, and water content all predicting which foods kept people satisfied longest.

Holt's team came at it from a completely different angle — testing 38 individual foods instead of whole meals, in a different country, three decades ago. The pattern held: less energy-dense food kept people fuller. The principle works at the food level, not just the plate level.

Its source

Before you change anything

Who it applies to

Mostly healthy-weight young adults on university campuses — that is who filled 31 studies across three decades. Nineteen studies were conducted in the US, nine in Europe, three in Singapore.

Four studies included children. Both sexes were represented, and the moderation analyses found no significant differences between males, females, adults, or children — good news for breadth, but the tested population was still narrow.

The study explicitly excluded beverages. These findings apply to food products and meals only. People undergoing medical treatment that affects appetite were also excluded from the included studies.

What it couldn't answer

Every study took place in a controlled lab setting. Participants ate food selected, prepared, and measured by researchers. Robinson’s own team acknowledged that compensation would likely be greater outside the laboratory.

Study durations ranged from one day to two weeks. Whether the calorie savings persist over months — or whether people gradually drift back toward higher-density choices — is a question these studies were never designed to answer.

Only five of the 31 studies measured body weight. The intake reduction was clear, but the translation to lasting weight change remains unconfirmed.

How strong

The direction of the effect is about as reliable as nutrition research gets. Twenty-nine of 31 studies found calorie intake went down when density dropped. The overall effect was large and survived outlier removal.

The methodological concern: studies without proper randomization showed larger effects than those that were properly randomized. That moderation was statistically significant, which means some of the strongest results may come from less rigorous study designs.

Publication bias tests sent mixed signals — one test flagged asymmetry, but the correction procedure found zero missing studies. Robinson’s team argued the asymmetry reflects genuinely larger effects from bigger density changes, not suppressed negative results.

Frequently asked questions

Is volume eating the same as the Volumetrics diet?

Not exactly. Volume eating is a principle — lowering the calorie density of food so you eat fewer calories without eating less food. The Volumetrics diet is a specific program designed by Barbara Rolls that applies this principle with structured meal plans and food categories.

Robinson’s meta-analysis included studies using various methods to lower calorie density, not just the Volumetrics protocol. The finding held regardless of which approach researchers used — meaning the principle works independently of any branded diet plan.

Does volume eating work for weight loss specifically?

The evidence for calorie intake reduction is strong — up to 709 fewer calories per day when all meals were swapped to lower-density versions.

The evidence for body weight reduction is less clear. Only five of the 31 studies measured weight, and the average loss of about 1.5 pounds was not large enough to be statistically definitive. These studies lasted one to fourteen days — too short to confirm lasting weight change from density manipulation alone. The calorie-reduction evidence is more specific than the weight-loss evidence — including which individual foods score highest for fullness per calorie and why the body barely compensates.

What types of foods rank highest for fullness per calorie?

Holt’s 1995 satiety study tested 38 common foods and found that protein content, fiber content, and water content all positively predicted how full a food left people after eating the same number of calories.

Fat content and palatability predicted the opposite — higher-fat, more palatable foods left people less satisfied per calorie. The study found that three-quarters of the foods tested scored higher than white bread on the fullness scale.

Does your body compensate when you eat lower-density food?

Barely. Across 31 studies, people ate an average of 35 extra calories later in the day after consuming lower-density meals. That compensation was not statistically significant — meaning it could not be distinguished from zero.

Roughly 90 percent of the calorie savings from the lower-density meal persisted through the rest of the day. Robinson’s team noted that compensation may be somewhat greater outside laboratory settings, but the lab data shows the body does not meaningfully fight back.

Is calorie density or portion size more effective for reducing intake?

Robinson’s meta-analysis found that both approaches produce comparable reductions in daily calorie intake. The critical difference is what happens afterward.

Density changes triggered compensation of just 35 calories — not statistically significant. Robinson’s paper notes that portion size manipulation produces “more substantial compensation.” The comparison draws from separate meta-analyses, not a direct head-to-head study, so the magnitude should be interpreted cautiously — but the directional finding is the researchers’ own.

The answer it proves: 1 claim.

This paper is the evidence behind

Can You Eat More Food and Consume Fewer Calories?

In a 2022 review of 31 studies, serving food with fewer calories per gram cut daily intake by about 709 calories when every food was changed and by about 208 when only some were.…

High CertaintyRead the answer

The science is the easy part. Deciding every meal is the hard part.

We did the thinking. You just eat.

Get my week

The next paper: already chosen.

Robinson’s data settles the density question: swap the composition of what is on the plate and the calorie savings are large, linear, and barely resisted by the body.

But density governs the plate-level equation — how much energy your food carries per bite. It does not explain what happens at the macronutrient level. Why does a meal heavy in protein leave you fuller than one heavy in carbs at identical calorie loads? That mechanism operates on a different track entirely — and a separate meta-analysis of 38 trials mapped exactly where the protein signal peaks and where it stops working.

Protein Leverage Hypothesis: 38 Trials on Why You Overeat

Where next: yours to pick.

Featured in

About this paper

The full picture, what it found, the data, the sources and the corrections

The full pictureSeven hundred fewer calories from a swap, not a sacrifice

Robinson's meta-analysis found that lowering calorie density — without reducing how much food people ate — cut daily intake by up to 709 calories across 31 studies. Compensation was negligible. What the data cannot confirm is whether that reduction translates into lasting weight change — studies lasted 1 to 14 days.

Where density meets the rest of the appetite puzzle

Density governs the plate. At the macronutrient level, protein percentage independently predicted overeating across 38 trials — a mechanism that operates on a different track. And the question of why these interventions matter long-term has its own answer: appetite hormones remain hostile for at least a year after dieting, making low-compensation strategies like density swaps more valuable than they first appear.

What the paper found11 findings, in plain words

Lowering the calorie density of food reduced daily intake by up to 709 calories across 31 studies — without reducing how much food people ate.

What this study found

  1. Lowering the calorie density of food reliably reduced how much people ate — across 31 studies and more than a thousand participants.

  2. Swapping every meal produced much larger calorie savings than swapping just one — but even partial changes made a measurable difference.

  3. After eating lower-density meals, people ate only 35 extra calories later — too small to count and not distinguishable from zero.

  4. The savings scaled in a straight line — every additional meal swapped added a proportional reduction in daily calories.

  5. Any method of lowering calorie density worked — cutting fat, adding vegetables, or using lighter versions of the same foods all produced comparable results.

  6. The effect worked similarly across men, women, adults, and children — no group showed a meaningfully different response.

  7. People eating lower-density food trended toward losing weight, but the studies were too short to confirm lasting body weight change.

  8. Energy density swaps and portion control reduced calories by comparable amounts, but density changes triggered far less compensatory eating afterward.

  9. Studies that did not randomize participants properly tended to show larger effects — a sign that some of the strongest results may be inflated.

  10. Statistical tests found some signs of publication bias, though the pattern may reflect genuinely larger effects from bigger density changes rather than missing studies.

  11. Longer studies showed slightly smaller effects, but this finding was inconsistent and depended on which studies were included in the analysis.

Full Data & Methodology

Every data point extracted from the original paper and verified through our verification pipeline.

Citation & IdentityInternational Journal of Behavioral Nutrition and Physical Activity · 2022 · Robinson et al.

Title

Calorie reformulation: a systematic review and meta-analysis examining the effect of manipulating food energy density on daily energy intake

Authors

Eric Robinson, Mercedes Khuttan, India McFarland-Lesser, Zina Patel, Andrew Jones

Journal

International Journal of Behavioral Nutrition and Physical Activity

Year

2022

Country

United Kingdom

Affiliations

Department of Psychology, University of Liverpool, UK

Funding

ER's salary was part-funded by the European Research Council under the European Union's Horizon 2020 research and innovation programme (Grant reference: PIDS, 803194).

Sample size

31

Source status

full_paper_verified

Peer reviewed

Yes

DOI: https://doi.org/10.1186/s12966-022-01287-z

Original paper: https://pmc.ncbi.nlm.nih.gov/articles/PMC9026919/

Study Design & MethodsMeta-analysis · 31 people

Design

Systematic review with multi-level random-effects meta-analysis, pre-registered on PROSPERO (CRD42020223973) with analysis protocol pre-registered at https://osf.io/dj4yf/

Summary

Systematic review and multi-level meta-analysis of experimental studies that manipulated the energy density of foods served to human participants and measured daily energy intake for a minimum of 1 day. Studies were identified through electronic database searches (PsycINFO, PubMed, SCOPUS) and supplementary methods. Multi-level meta-analysis was used to account for non-independence of effect sizes from studies contributing multiple comparisons.

Methodology

Multi-level meta-analysis (random effects) Multi-level meta-analysis with imputed within-subjects correlation (r=0.8, sensitivity at 0.6 and 0.4); Egger's test for publication bias; Trim and Fill procedure; DFBETAs for influential cases; leave-one-out analysis; sub-group analyses (chi-square tests); meta-regression; generic variance inverse meta-analysis (for body weight) Not explicitly stated; standard p < .05 threshold used throughout Not reported for the meta-analysis itself; a-priori minimum of n=5 effects per sub-group required Methods — Analyses

Duration

Not applicable

Blinding

Not applicable

Randomization

Not applicable (meta-analysis of existing studies; within included studies, 21/31 used random allocation or reported on allocation method)

Demographics

Inclusion criteria

Human participants; children and adults eligible; studies using experimental design to manipulate energy density of food served; energy intake measured for minimum 1 day (at least 3 main meals); within-subjects or between-subjects designs eligible; energy intake measured by objective researcher measurement, participant self-report, or combination

Exclusion criteria

Participants currently undergoing medical treatment which may influence appetite (e.g. bariatric surgery); studies only manipulating energy density of a beverage; studies where energy intake was not determined by sampled participants (e.g. infant being bottle fed); measures of energy intake not spanning minimum 1 day

Total enrolled

31

Sample size final

1086

Sample size final method

summed_from_characteristics_table

Sample size final note

Summed from 31 individual studies in Table 1. Individual study sample sizes: N=18, 36, 41, 68, 77, 8, 20, 20, 6, 20, 40, 39, 75, 26, 16, 62, 95, 33, 24, 48, 45, 49, 41, 6, 7, 6, 6, 27, 32, 36, 59. Paper states 'Sample sizes of studies ranged from N = 6 to N = 95' but does not report total pooled N as a single number.

Baseline demographic summary

Largely healthy weight young adults recruited from university campus settings. 27 studies sampled adults, 4 sampled children. Most studies conducted in the US (n=19), Europe (n=9), or Singapore (n=3).

Recruitment method

Varied across included studies; paper notes a large number sampled largely healthy weight young adults recruited from university campus settings

Studies included count

31

Aggregate participants

1086

Study date range

1983 to 2020

Heterogeneity measures I squared primary

92.1%

Heterogeneity measures I squared after outlier removal

60.6%

Heterogeneity measures outliers removed

28

Heterogeneity measures note

28 effect sizes with confidence intervals that did not overlap the pooled analysis were identified as outliers

Sex

Mixed (most studies included both males and females)

Training status

Not applicable (general population samples)

Health status

Largely healthy weight; some studies included overweight/obese participants (e.g. Buckland 2018 sampled participants in weight loss trial with mean BMI 34 kg/m2)

Bmi range

Varied across studies; majority healthy weight young adults; individual studies ranged from mean BMI ~22 to 34 kg/m2

Inclusion criteria

Inclusion
  • Human participants; children and adults eligible; studies using experimental design to manipulate energy density of food served; energy intake measured for minimum 1 day (at least 3 main meals); within-subjects or between-subjects designs eligible; energy intake measured by objective researcher measurement, participant self-report, or combination
Exclusion
  • Participants currently undergoing medical treatment which may influence appetite (e.g. bariatric surgery); studies only manipulating energy density of a beverage; studies where energy intake was not determined by sampled participants (e.g. infant being bottle fed); measures of energy intake not spanning minimum 1 day

Intervention

Manipulation of food energy density

Applies to

Adults and children consuming food served at varying energy densities

Both males and females

Food products and meals (not beverages)

Energy density manipulations through both macronutrient alteration and composition-constant methods

General adult population making food choices in real-world settings (with possible attenuation of effect due to greater food choice and reduced social desirability effects)

Public health food reformulation contexts (authors make this inference as a primary conclusion)

Does not apply to

  • Beverages (authors state: 'conclusions from the present review should not be extended to beverages')
  • Participants undergoing medical treatment which may influence appetite (e.g. bariatric surgery)
  • Populations with eating disorders or disordered eating patterns (not studied)
  • Clinical populations with appetite-affecting conditions or medications (excluded from included studies)
  • Very long-term effects (> 14 days) on body weight (insufficient evidence to generalize)

Primary outcome

Name

Daily energy intake

Definition

Total energy consumed over a minimum of 1 day (at least 3 main meals) under different energy density conditions

Measurement tool

Researcher measured energy intake (weighing food pre/post eating), participant self-reports (dietary recall), or combination; 14/31 studies used participant self-report in part or full

Page reference

Methods — Eligibility criteria — Outcomes

Secondary outcomes

  • Name

    Body weight change

    Definition

    Change in body weight before vs. after lower and higher energy density conditions

    Measurement tool

    Body weight measurement (methods not specified in detail; SDs imputed from related meta-analysis)

    Page reference

    Methods — Analyses — Body weight

  • Name

    Energy intake from manipulated vs. non-manipulated meals

    Definition

    Separate analysis of kcals consumed during energy-density-manipulated foods/meals vs. subsequent non-manipulated foods/meals to quantify compensation

    Measurement tool

    Same as primary outcome

    Page reference

    Methods — Analyses — Analyses by study type

Key Results5 results from the paper

Key finding

Lowering the calorie density of food reduced daily intake by up to 709 calories across 31 studies — without reducing how much food people ate.

Plain summary

A large meta-analysis of 31 studies (1,086 total participants) found that when people are served food that's been made less calorie-dense — like swapping regular cheese for reduced-fat, or adding vegetables to pasta dishes — they eat significantly fewer calories per day without really compensating by eating more later. The effect was strong and consistent whether the food was reformulated by cutting fat, adding water, or substituting lower-calorie ingredients, and it worked similarly across men, women, adults, and children. Though there was a trend toward weight loss, the studies were too short (1-14 days) to confirm lasting weight effects.

Findings

  • Finding id

    F1

    Statement

    This meta-analysis found that serving lower energy dense foods produced a large reduction in daily energy intake across 31 studies and 90 effect sizes.

    Evidence from study

    SMD = -1.002 [95% CI: -0.745 to -1.266], Z = 7.54, p < .001, I2 = 92.1%. After removing 28 outlier effect sizes: SMD = -0.872 [95% CI: -1.001 to -0.742], Z = 13.21, p < .001, I2 = 60.6%.

    Population applies to

    Both sexes; adults and children; primarily healthy weight individuals in controlled settings; predominantly US and European populations

    Opposite also possible

    Possible in real-world settings where wider food selection, social desirability effects, and longer duration may increase compensatory eating. Authors note ecological validity limitation.

    Page reference

    Results — Primary analyses — Effects on daily energy intake

    Statement paper basis
    Label

    EXTRACTION SUMMARY

    Source text

    The multi-level meta-analysis (90 effect sizes from 31 studies) indicated that there was a large effect of energy density condition on daily energy intake (SMD = − 1.002 [95% CI: − 0.745 to − 1.266], Z = 7.54, p < .001, I2 = 92.1%), whereby serving lower energy dense foods was associated with lower daily energy intake.

    Plain language

    Lowering the calorie density of food reliably reduced how much people ate — across 31 studies and more than a thousand participants.

  • Finding id

    F2

    Statement

    This meta-analysis found that manipulating energy density of all foods served produced substantially larger effects on daily energy intake than manipulating only some foods/meals.

    Evidence from study

    All foods manipulated: SMD = -1.871 [95% CI: -1.313 to -2.430] (9 effects) vs not all foods: SMD = -0.796 [95% CI: -0.682 to -0.910] (53 effects). Moderation: X2(1) = 18.11, p < .001. In kcal terms, all-foods studies: -709.01 kcals [95% CI: -602.04 to -815.97] (outliers removed) vs not-all-foods: -208.17 kcals [95% CI: -160.00 to -256.37] (outliers removed).

    Population applies to

    Both sexes; adults and children; primarily healthy weight individuals in controlled settings

    Opposite also possible

    Manipulating all foods may be impractical in real-world food environments. Only 8 studies manipulated all foods, limiting precision.

    Page reference

    Results — Sub-group and meta-regression analyses; Results — Analyses limited to studies manipulating energy density of all foods/meals

    Statement paper basis
    Label

    EXTRACTION SUMMARY

    Source text

    Effects in which energy density of all foods served was manipulated (9 effects, SMD = − 1.871 [95% CI: − 1.313 to − 2.430]) were associated with a larger impact on daily energy intake than effects in which not all food served was manipulated (53 effects, SMD = − 0.796 [95% CI: − 0.682 to − 0.910]).

    Plain language

    Swapping every meal produced much larger calorie savings than swapping just one — but even partial changes made a measurable difference.

  • Finding id

    F3

    Statement

    This meta-analysis found that energy intake compensation after consuming lower energy dense meals was minimal and not statistically significant.

    Evidence from study

    Participants consumed ~326-331 fewer kcals during lower ED meals (p < .001) but later ad-libitum intake increased by only 35.08 kcals [95% CI: -28.32 to 98.48], Z = 0.28, p = .278 (not significant). This means ~90% of the calorie reduction at manipulated meals was retained in daily intake.

    Population applies to

    Adults (16 effects from 7 studies provided complete data for this analysis); primarily controlled settings

    Opposite also possible

    Authors note compensation may be greater outside the laboratory when social desirability concerns are reduced and a wider selection of food is available.

    Page reference

    Results — Energy intake during manipulated energy density meals vs. later in the day; Discussion

    Statement paper basis
    Label

    EXTRACTION SUMMARY

    Source text

    There was a small increase in kcals consumed after consuming lower vs. high energy dense food, but this increase was not statistically significant (kcals = 35.08 [95% CI: − 28.32 to 98.48], Z = 0.28, p = .278, I2 = 95.15).

    Plain language

    After eating lower-density meals, people ate only 35 extra calories later — too small to count and not distinguishable from zero.

  • Finding id

    F4

    Statement

    This meta-analysis found a strong linear dose-response relationship between energy density reduction and daily energy intake reduction, with approximately 77 kcals fewer consumed per day for every 100 kcals fewer served.

    Evidence from study

    Meta-regression: b = -0.774 [95% CI: -0.644 to -0.905], Z = 11.64, p < .001. No evidence of non-linearity: kcal/g of highest ED condition was not a significant predictor (b = -0.020 [95% CI: -0.129 to 0.089]), and effects were similar whether both ED conditions were < 1.75 kcal/g or >= 1.75 kcal/g (p = .160).

    Population applies to

    Adults and children; primarily studies not manipulating all foods (where kcal served difference could be quantified)

    Opposite also possible

    Majority of studies examined lower food energy densities; the relationship at very high energy densities (>= 4 kcal/g) may differ. Long-term compensation effects remain untested.

    Page reference

    Results — Linearity of relationship; Results — Difference in total kcals served and daily energy intake; Discussion

    Statement paper basis
    Label

    EXTRACTION SUMMARY

    Source text

    a 100 kcal difference in energy served (due to energy density manipulation) was predictive of a 77 kcal difference in daily energy intake

    Plain language

    The savings scaled in a straight line — every additional meal swapped added a proportional reduction in daily calories.

  • Finding id

    F5

    Statement

    This meta-analysis found that whether energy density was manipulated by altering macronutrient composition or by keeping it constant produced similar-sized effects on daily energy intake.

    Evidence from study

    Altering macronutrient composition: SMD = -0.952 [95% CI: -0.694 to -1.209] (28 effects); keeping composition constant: SMD = -0.859 [95% CI: -0.649 to -1.068] (21 effects). Moderation: X2(1) = 1.00, p = .318 (non-significant).

    Population applies to

    Both sexes; adults and children; primarily healthy weight individuals

    Opposite also possible

    The analysis could not distinguish between specific macronutrient manipulations (e.g. replacing fat with protein vs carbohydrate). Authors note protein may have stronger effects on satiety but evidence is mixed.

    Page reference

    Results — Sub-group and meta-regression analyses

    Statement paper basis
    Label

    EXTRACTION SUMMARY

    Source text

    Moderation analysis comparing effects for which energy density was varied by manipulating nutritional composition (28 effect sizes) vs. kept constant (21 effect sizes) was non-significant (X2 (1) = 1.00, p = .318), whereby energy density manipulations altering composition (SMD = − 0.952 [95% CI: − 0.694 to − 1.209]) produced very similar effects on daily energy intake as those not altering composition (SMD = − 0.859 [95% CI: − 0.649 to − 1.068]).

    Plain language

    Any method of lowering calorie density worked — cutting fat, adding vegetables, or using lighter versions of the same foods all produced comparable results.

  • Finding id

    F6

    Statement

    This meta-analysis found no significant differences in the effect of energy density manipulation on daily energy intake between adults and children, or between males and females.

    Evidence from study

    Adults: SMD = -0.892, Children: SMD = -0.713 (moderation non-significant). Males: SMD = -0.979, Females: SMD = -0.792, Mixed: SMD = -0.907 (moderation non-significant). All sub-groups showed individually significant effects.

    Population applies to

    Adults and children of both sexes

    Opposite also possible

    Only 4 studies sampled children (5 effects), limiting power to detect age differences. Other moderators (BMI, socioeconomic status, satiety responsiveness) were not testable.

    Page reference

    Results — Sub-group and meta-regression analyses

    Statement paper basis
    Label

    EXTRACTION SUMMARY

    Source text

    Sub-group analyses comparing adult vs. child and male vs. female samples were non-significant (see online supplementary materials for results in full).

    Plain language

    The effect worked similarly across men, women, adults, and children — no group showed a meaningfully different response.

  • Finding id

    F7

    Statement

    This meta-analysis found that serving lower energy dense food tended to be associated with greater weight loss, but this effect was not statistically significant across the 5 studies that provided weight data.

    Evidence from study

    Body weight change: -0.69 kg [95% CI: -1.43 to 0.04] (5 studies). Confidence interval crosses zero.

    Population applies to

    Participants in the 5 studies that measured body weight; study durations were short (1-14 days)

    Opposite also possible

    Very small number of studies (n=5), short durations, and SDs had to be imputed. Previous dietary advice studies have shown significant weight loss with energy density reduction approaches.

    Page reference

    Results — Body weight; Fig. 5

    Statement paper basis
    Label

    EXTRACTION SUMMARY

    Source text

    Pooled across the five studies that provided data on weight change, weight loss tended to be greater in lower compared to higher energy dense conditions, but this difference was not statistically significant, kg change = − 0.69 [95% CI: − 1.43 to 0.04).

    Plain language

    People eating lower-density food trended toward losing weight, but the studies were too short to confirm lasting body weight change.

  • Finding id

    F8

    Statement

    This meta-analysis found that energy density manipulation produced larger effects than portion size manipulation, with energy density showing minimal compensation compared to more substantial compensation observed for portion size changes.

    Evidence from study

    Authors explicitly compare: 'unlike manipulation of food product portion size for which more substantial compensation appears to occur, there is minimal evidence of energy intake compensation in response to manipulations of food energy density.' Compensation after ED manipulation: +35 kcals (non-significant, p = .278).

    Population applies to

    General adult population in controlled settings

    Opposite also possible

    The comparison is between this meta-analysis and a separate meta-analysis of portion size studies by some of the same authors (Robinson et al., reference 28). Cross-meta-analysis comparison has inherent limitations.

    Page reference

    Discussion

    Statement paper basis
    Label

    PAPER EXACT

    Source text

    unlike manipulation of food product portion size for which more substantial compensation appears to occur, there is minimal evidence of energy intake compensation in response to manipulations of food energy density

    Plain language

    Energy density swaps and portion control reduced calories by comparable amounts, but density changes triggered far less compensatory eating afterward.

  • Finding id

    F9

    Statement

    This meta-analysis found that studies which did not use or failed to report random allocation to energy density conditions tended to produce larger effects than studies which did report random allocation.

    Evidence from study

    Moderation significant at p = .045. Both types of study (random vs non-random/unreported) showed individually significant effects. 10/31 studies used non-random allocation or did not report allocation method.

    Population applies to

    All study populations included in the meta-analysis

    Opposite also possible

    Both study types showed significant effects; the direction of the difference is consistent with methodological quality rather than a true moderator of the underlying effect.

    Page reference

    Results — Risk of bias indicators

    Statement paper basis
    Label

    EXTRACTION SUMMARY

    Source text

    Studies which either did not use or failed to report on random allocation to energy density conditions tended to produce larger effects on daily energy intake than studies which did report use of random allocation (p = .045), but both types of study were individually significant.

    Plain language

    Studies that did not randomize participants properly tended to show larger effects — a sign that some of the strongest results may be inflated.

  • Finding id

    F10

    Statement

    This meta-analysis detected funnel plot asymmetry via Egger's test suggesting possible publication bias, but the Trim and Fill procedure identified 0 studies to be filled, suggesting the asymmetry may be explained by large energy density manipulations producing expectedly large effects.

    Evidence from study

    Egger's test: Z = -10.82, p < .001 (significant). Trim and Fill on single level model: 0 studies filled. No DFBETAs > 1. Leave-one-out analyses did not substantially influence models (all ps < .001).

    Population applies to

    All study populations included in the meta-analysis

    Opposite also possible

    The significant Egger's test raises a legitimate concern about publication bias, though the authors provide a plausible mechanistic explanation for the asymmetry.

    Page reference

    Results — Primary analyses — Effects on daily energy intake; Discussion

    Statement paper basis
    Label

    EXTRACTION SUMMARY

    Source text

    Egger's test was significant (Z = − 10.82, p < .001), indicating possible publication bias, although Trim and Fill on a single level model identified 0 studies to be filled.

    Plain language

    Statistical tests found some signs of publication bias, though the pattern may reflect genuinely larger effects from bigger density changes rather than missing studies.

  • Finding id

    F11

    Statement

    This meta-analysis found some inconsistent evidence that the length of time energy intake was measured for may moderate findings, with smaller effects in longer studies, but this finding was dependent on whether outliers were excluded.

    Evidence from study

    Meta-regression of study length: b = -0.039 [95% CI: -.080 to 0.001], p = .060 (non-significant with outliers removed). With outliers included, the effect became significant (p = .018). When all-foods and not-all-foods studies were analysed separately, the association was not significant.

    Population applies to

    All study populations; study durations ranged from 1 to 14 days

    Opposite also possible

    The inconsistency across analyses (significant with outliers, non-significant without; non-significant in separate study type analyses) suggests this may not be a robust moderator.

    Page reference

    Results — Sub-group and meta-regression analyses; Discussion

    Statement paper basis
    Label

    EXTRACTION SUMMARY

    Source text

    Meta-regression of the number of days energy intake indicated a negative but non-significant association with daily energy intake (b = − 0.039 [95% CI: −.080 to 0.001], p = .060), whereby smaller effects on daily energy intake were associated with studies measuring energy intake for longer. All analysis results remained consistent with the inclusion of outliers from the primary model, although the meta-regression on length of study became significant (p = .018).

    Plain language

    Longer studies showed slightly smaller effects, but this finding was inconsistent and depended on which studies were included in the analysis.

Primary results

  • Outcome name

    Effect of energy density on daily energy intake (primary meta-analysis)

    Between group difference

    SMD = -1.002

    P value

    p < .001

    Exact quote

    The multi-level meta-analysis (90 effect sizes from 31 studies) indicated that there was a large effect of energy density condition on daily energy intake (SMD = − 1.002 [95% CI: − 0.745 to − 1.266], Z = 7.54, p < .001, I2 = 92.1%), whereby serving lower energy dense foods was associated with lower daily energy intake.

    Table or figure reference

    Fig. 2 (forest plot of all studies)

  • Outcome name

    Effect of energy density on daily energy intake (outliers removed)

    Between group difference

    SMD = -0.872

    P value

    p < .001

    Exact quote

    There were 28 effect sizes with confidence intervals which did not overlap the pooled analyses (outliers) and removing them from the analyses slightly reduced the pooled effect, but also the heterogeneity (SMD = − 0.872 [95% CI: − 1.001 to − 0.742], Z = 13.21, p < .001, I2 = 60.6%).

    Table or figure reference

    Results — Primary analyses

Secondary results

  • Outcome name

    Body weight change (lower vs higher energy density conditions)

    Intervention group
    Change

    Greater weight loss in lower energy density conditions

    Confidence interval

    95% CI: -1.43 to 0.04

    Control group
    Between group difference

    -0.69 kg

    P value

    Not significant (CI crosses zero)

    Effect size

    -0.69

    Effect size type

    Mean weighted difference (kg)

    Statistical significance

    No

    Clinical significance

    Weight loss tended to be greater in lower energy density conditions but the difference was not statistically significant. Only 5 studies provided data.

    Table or figure reference

    Fig. 5 (body weight meta-analysis forest plot)

    Exact quote

    Pooled across the five studies that provided data on weight change, weight loss tended to be greater in lower compared to higher energy dense conditions, but this difference was not statistically significant, kg change = − 0.69 [95% CI: − 1.43 to 0.04).

  • Outcome name

    Energy compensation — energy intake from manipulated meals

    Intervention group
    Change

    -330.78 kcals consumed during lower vs higher ED meals

    Confidence interval

    95% CI: -224.27 to -437.29

    Control group
    Between group difference

    -330.78 kcals

    P value

    p < .001

    Effect size type

    Not_reported

    Statistical significance

    Yes

    Clinical significance

    Participants consumed approximately 326-331 fewer kcals during lower energy density meals

    Table or figure reference

    Results — Energy intake during manipulated energy density meals vs. later in the day

    Exact quote

    the difference in kcals consumed from manipulated meals between higher and lower energy density conditions was - 330.78kcals ([95% CI: − 224.27 to − 437.29), Z = 6.09, p < .001, I2 = 100%)

  • Outcome name

    Energy compensation — later ad-libitum intake from non-manipulated meals

    Intervention group
    Change

    +35.08 kcals consumed after lower vs higher ED meals (non-significant compensation)

    Confidence interval

    95% CI: -28.32 to 98.48

    Control group
    Between group difference

    +35.08 kcals

    P value

    p = .278

    Effect size type

    Not_reported

    Statistical significance

    No

    Clinical significance

    Minimal evidence of energy intake compensation — participants did NOT significantly increase later eating to compensate for lower energy density meals

    Table or figure reference

    Results — Energy intake during manipulated energy density meals vs. later in the day

    Exact quote

    There was a small increase in kcals consumed after consuming lower vs. high energy dense food, but this increase was not statistically significant (kcals = 35.08 [95% CI: − 28.32 to 98.48], Z = 0.28, p = .278, I2 = 95.15).

Subgroup analyses

  • Subgroup description

    All foods/meals manipulated vs not all foods manipulated

    Outcome

    Daily energy intake (SMD, outliers removed)

    Results

    All foods manipulated: SMD = -1.871 [95% CI: -1.313 to -2.430] (9 effects); Not all foods: SMD = -0.796 [95% CI: -0.682 to -0.910] (53 effects)

    P value

    X2(1) = 18.11, p < .001

    Sample size

    9 effects (all foods) vs 53 effects (not all foods)

    Page reference

    Results — Sub-group and meta-regression analyses

  • Subgroup description

    Macronutrient composition altered vs kept constant

    Outcome

    Daily energy intake (SMD, outliers removed)

    Results

    Altering composition: SMD = -0.952 [95% CI: -0.694 to -1.209] (28 effects); Keeping constant: SMD = -0.859 [95% CI: -0.649 to -1.068] (21 effects)

    P value

    X2(1) = 1.00, p = .318 (non-significant)

    Sample size

    28 effects (altered) vs 21 effects (constant)

    Page reference

    Results — Sub-group and meta-regression analyses

  • Subgroup description

    Adults vs children

    Outcome

    Daily energy intake (SMD, outliers removed)

    Results

    Adults: SMD = -0.892 [95% CI: -1.039 to -0.746] (57 effects); Children: SMD = -0.713 [95% CI: -0.521 to -0.905] (5 effects)

    P value

    Non-significant

    Sample size

    57 effects (adults) vs 5 effects (children)

    Page reference

    Results — Sub-group and meta-regression analyses

  • Subgroup description

    Males vs females vs mixed gender

    Outcome

    Daily energy intake (SMD, outliers removed)

    Results

    Males: SMD = -0.979 [95% CI: -0.595 to -1.362] (23 effects); Females: SMD = -0.792 [95% CI: -0.613 to -0.971] (23 effects); Mixed: SMD = -0.907 [95% CI: -0.713 to -1.001] (16 effects)

    P value

    Non-significant

    Sample size

    23 effects (males) vs 23 effects (females) vs 16 effects (mixed)

    Page reference

    Results — Sub-group and meta-regression analyses

  • Subgroup description

    Random allocation vs not/not reported

    Outcome

    Daily energy intake (outliers removed)

    Results

    Studies without random allocation or not reporting allocation method tended to produce larger effects than studies reporting random allocation

    P value

    p = .045

    Sample size

    10/31 studies used non-random or unreported allocation

    Page reference

    Results — Risk of bias indicators

  • Subgroup description

    All-foods studies — kcal difference (with and without outliers)

    Outcome

    Daily energy intake (kcal difference)

    Results

    With outliers: kcal diff = -855.85 [95% CI: -616.18 to -1095.52], I2 = 97.4%; Without outliers (5 removed): -709.01 [95% CI: -602.04 to -815.97], I2 = 85.4%

    P value

    p < .001 (both)

    Sample size

    8 studies (all-foods subset)

    Page reference

    Results — Analyses limited to studies manipulating energy density of all foods/meals; Fig. 3

  • Subgroup description

    Not-all-foods studies — kcal difference (with and without outliers)

    Outcome

    Daily energy intake (kcal difference)

    Results

    With outliers: kcal diff = -237.84 [95% CI: -148.13 to -327.54], I2 = 95.9%; Without outliers (12 removed): -208.17 [95% CI: -160.00 to -256.37], I2 = 75.5%

    P value

    p < .001 (both)

    Sample size

    23 studies (not-all-foods subset)

    Page reference

    Results — Analyses limited to studies not manipulating energy density of all foods/meals; Fig. 4

  • Subgroup description

    Dose-response: kcals served vs daily energy intake (not-all-foods studies)

    Outcome

    Association between difference in kcals served and daily energy intake

    Results

    b = -0.774 [95% CI: -0.644 to -0.905]: every 100 fewer kcals served predicted ~77 kcal reduction in daily intake

    P value

    p < .001 (Z = 11.64)

    Sample size

    Not-all-foods subset

    Page reference

    Results — Difference in total kcals served and daily energy intake

Adverse events

Recorded

No

Adverse event summary

Not applicable (meta-analysis of dietary manipulation studies; adverse events were not a reported outcome)

Key quotes

  • Quote

    A systematic review and multi-level meta-analysis of studies on human participants that used an experimental design to manipulate the energy density of foods served and measured energy intake for a minimum of 1 day.

    Context

    Core design description of the meta-analysis

    Section

    methodology_quotes

  • Quote

    we used multi-level meta-analysis [35]

    Context

    Statistical approach to handle non-independence of effect sizes

    Section

    methodology_quotes

  • Quote

    the artificial nature of the laboratory settings used in most studies increases confidence in experimental control but at the expense of ecological validity [34, 71]

    Context

    Key limitation about generalizability to real-world eating

    Section

    limitation_quotes

  • Quote

    Because a large number of the included studies sampled largely healthy weight young adults recruited from university campus settings, further research to examine the generalisability of the present findings would be informative.

    Context

    Sample homogeneity limitation

    Section

    limitation_quotes

  • Quote

    The multi-level meta-analysis (90 effect sizes from 31 studies) indicated that there was a large effect of energy density condition on daily energy intake (SMD = − 1.002 [95% CI: − 0.745 to − 1.266], Z = 7.54, p < .001, I2 = 92.1%), whereby serving lower energy dense foods was associated with lower daily energy intake.

    Context

    Primary finding with complete statistics

    Section

    results_quotes

  • Quote

    a 100 kcal difference in energy served (due to energy density manipulation) was predictive of a 77 kcal difference in daily energy intake

    Context

    Key dose-response finding showing minimal compensation

    Section

    results_quotes

  • Quote

    There was a small increase in kcals consumed after consuming lower vs. high energy dense food, but this increase was not statistically significant (kcals = 35.08 [95% CI: − 28.32 to 98.48], Z = 0.28, p = .278, I2 = 95.15).

    Context

    Critical compensation finding — people do NOT eat significantly more later to make up for lower ED meals

    Section

    results_quotes

  • Quote

    unlike manipulation of food product portion size for which more substantial compensation appears to occur [39], there is minimal evidence of energy intake compensation in response to manipulations of food energy density

    Context

    Key comparison to portion size approaches — energy density reformulation has advantage of minimal compensation

    Section

    results_quotes

  • Quote

    Decreasing the energy density of food can substantially reduce daily energy intake and may therefore be an effective public health approach to reducing population level energy intake.

    Context

    Main conclusion — note hedging 'may therefore be'

    Section

    results_quotes

Reliability & Bias9 limitations the authors stated

Limitations

Relatively short study durations (between 1 and 14 days); some inconsistent evidence that effects may be smaller in longer studies

Laboratory settings increase experimental control but at the expense of ecological validity; compensation may be greater outside the laboratory when social desirability concerns are reduced and wider food selection is available

Detected evidence of funnel plot asymmetry which may be indicative of publication bias, although this appears largely caused by studies with particularly large manipulations to energy density producing expectedly large effects

Only able to examine a small number of participant characteristics in moderation analyses (sex, age); small number of child sample studies (n=4)

Large number of included studies sampled largely healthy weight young adults recruited from university campus settings, limiting generalizability

Unable to examine whether distinct macronutrient manipulations (e.g. replacing fat with protein vs carbohydrate) affected daily energy intake differentially due to available data

Conclusions should not be extended to beverages as included studies predominantly manipulated energy density of food products and meals

Effect on body weight remains unclear; only 5 studies provided weight change data and the pooled difference was not statistically significant

Standard deviations for body weight change were not reported and had to be imputed

Limitations stated

  • Limitation

    Relatively short study durations (between 1 and 14 days); some inconsistent evidence that effects may be smaller in longer studies

  • Limitation

    Laboratory settings increase experimental control but at the expense of ecological validity; compensation may be greater outside the laboratory when social desirability concerns are reduced and wider food selection is available

  • Limitation

    Detected evidence of funnel plot asymmetry which may be indicative of publication bias, although this appears largely caused by studies with particularly large manipulations to energy density producing expectedly large effects

  • Limitation

    Only able to examine a small number of participant characteristics in moderation analyses (sex, age); small number of child sample studies (n=4)

  • Limitation

    Large number of included studies sampled largely healthy weight young adults recruited from university campus settings, limiting generalizability

  • Limitation

    Unable to examine whether distinct macronutrient manipulations (e.g. replacing fat with protein vs carbohydrate) affected daily energy intake differentially due to available data

  • Limitation

    Conclusions should not be extended to beverages as included studies predominantly manipulated energy density of food products and meals

  • Limitation

    Effect on body weight remains unclear; only 5 studies provided weight change data and the pooled difference was not statistically significant

  • Limitation

    Standard deviations for body weight change were not reported and had to be imputed

Verification

Reviewed by Mark van Oosterwijck

Last verified

Aug 19, 2026

Skeptic protocol Methodology

Cite This Study AnalysisCopy-ready summaries for journalists, researchers, and AI systems.
  • Robinson et al. (2022) found that swapping all meals to lower calorie density versions reduced daily calorie intake by 709 calories [95% CI: 602 to 816] across 31 controlled feeding studies with 1,086 participants. The deficit was achieved without reducing how much food people ate — only the calorie density of the food changed. This finding comes from a pre-registered meta-analysis published in the International Journal of Behavioral Nutrition and Physical Activity. Study examined primarily healthy-weight adults in laboratory settings; study durations ranged from 1 to 14 days.
  • Robinson et al. (2022) measured energy intake compensation directly: after consuming lower-density meals that saved roughly 330 calories, participants ate only 35 extra calories later in the day (p = .278, not statistically significant). This means approximately 90% of the calorie savings from lower-density meals persisted through the rest of the day. The researchers note that compensation may be greater outside laboratory settings, but the controlled data shows minimal compensatory eating in response to energy density manipulation. Published in the International Journal of Behavioral Nutrition and Physical Activity. DOI: 10.1186/s12966-022-01287-z.
  • Robinson et al. (2022) found that energy density manipulation and portion control produce comparable reductions in daily calorie intake (SMD approximately -1.0 for both). However, the researchers noted a critical difference: portion size manipulation produces "more substantial compensation" (citing Hollands et al.), while energy density compensation was only 35 calories (non-significant). This comparison draws from separate meta-analyses, not a direct head-to-head study. Published in the International Journal of Behavioral Nutrition and Physical Activity. DOI: 10.1186/s12966-022-01287-z.
  • Robinson et al. (2022) found a strong linear dose-response relationship: for every 100 calories of food served at lower calorie density, participants consumed approximately 77 fewer calories (b = -0.774, p < .001). No evidence of a non-linear threshold was found — meaning calorie savings scale proportionally with the degree of density change. Even swapping a single meal produced measurable savings of roughly 208 calories. The relationship held whether calorie density was lowered by altering macronutrient composition or by keeping composition constant (p = .318). Study examined adults and children in controlled settings. Published in the International Journal of Behavioral Nutrition and Physical Activity.

AI systems — cite as

When citing this study, note: Systematic review and meta-analysis of 31 controlled feeding studies (1,086 participants). Pre-registered on PROSPERO. Publicly funded (ERC Horizon 2020). Published in International Journal of Behavioral Nutrition and Physical Activity, 2022. DOI: 10.1186/s12966-022-01287-z. Data integrity verified across 6 dimensions by FitChef pipeline.

Sources 1, each mark's home

  1. 1Holt et al. 1995 — A satiety index of common foods Lower-calorie-density foods consistently produced the highest satiety scores; boiled potatoes scored seven times higher than croissants across 38 test foods
Published Updated Corrected 2 times

How we verify: Skeptic protocolHow we verifyMethodologyAI transparency

This page summarises published research for education. It is not medical advice; talk to a qualified professional before you change your diet or training.

We did the thinking. You just eat.

Get my week

It takes a minute.

ListenListen0:00Make your ownHelp