Ketogenic Diets Are Not Beneficial for Athletic Performance: Response to Noakes

This is a meta-analysis and critical review that examined the effect of ketogenic low-carbohydrate high-fat (LCHF) diets on endurance sports performance. The researchers compared two dietary approaches:

**Ketogenic LCHF diet:** Extreme carbohydrate restriction (<50 grams per day), high fat intake, designed to induce nutritional ketosis

**High-carbohydrate availability (HCHO) diet:** Standard or high carbohydrate intake, typical of conventional sports nutrition

The primary outcome was **change in endurance performance**, measured as either:

Time trial performance (time to complete a set distance)

Time to exhaustion (how long an athlete could exercise at a fixed intensity)

Average power output during a set effort

The authors defined a **2% change in performance** as the threshold for real-world significance, based on doubling the typical 1% within-athlete variation in competitive events.

The analysis included data from **19 separate data points** across multiple studies, involving:

**Total participants:** Approximately 100–120 competitive endurance athletes across all studies (exact total not pooled, but individual study samples ranged from n=1 to n=19)

**Population:** Tier 2 to Tier 5 athletes (trained to elite/competitive level) including:

- Racewalkers (male and female)

- Cyclists (male)

- Runners (male, middle-aged and younger)

- Triathletes (male)

- One elite male triathlete (case study)

**Setting:** Mix of field conditions (real-life 10,000-m track races, 20-km road races) and laboratory conditions (cycling ergometers, treadmill time trials)

**Duration of LCHF interventions:** Ranged from 4 days to 2 years, with most studies using 3–4 weeks of dietary adaptation

**Sex:** Predominantly male; only 2 studies included female athletes (with no control for menstrual phase)

The authors used a **dashboard approach** rather than a traditional meta-analytical pooling of data. For each study, they extracted:

**Performance outcomes:** Published mean differences between LCHF and HCHO conditions for time trial time, time to exhaustion, or average power

**Data extraction:** From published tables or by digitalizing figures using plotdigitizer.com

**Conversion:** Time-to-exhaustion protocols were converted to estimated time trial performance changes using the methods of Hopkins et al.

**Context and caveats:** Each study was rated for:

- Study design (parallel-group vs. crossover)

- Randomization status (randomized vs. non-randomized/belief-chosen allocation)

- Dietary control (fully supervised vs. ad libitum with education)

- Presence of confounds (weight loss, order effects, suboptimal nutrition in control conditions)

- Performance validity (e.g., treadmill testing while measuring expired gases may limit pacing)

**Study design:** This is a **meta-analysis** with a unique "dashboard" approach. Rather than statistically pooling all results into a single effect size, the authors presented each study individually with its context and caveats, then synthesized qualitatively.

**Key design features:**

**Inclusion criteria:** Only studies of ketogenic diets (<50 g carbohydrate/day), in humans, with habitually trained endurance athletes (Tier 2 or above), using protocols relevant to endurance sports performance

**Comparison:** Each LCHF intervention was compared to a high-carbohydrate availability condition (HCHO or control)

**Performance threshold:** A 2% change was set as the minimum for real-world significance

**Analysis:** Difference between means calculated for each test; for parallel-group designs, this was the difference in change scores between groups; for crossover designs, direct comparison between treatments

**What this design can prove:**

This approach can identify **patterns across studies** — whether LCHF consistently improves, harms, or has no effect on performance

The dashboard method preserves **individual study nuances** that traditional meta-analysis might obscure (e.g., the role of weight loss, order effects, or non-randomization)

It can highlight **which study designs produce positive results** and whether those results hold up under scrutiny

**What this design cannot prove:**

It cannot produce a **single pooled effect size** with confidence intervals, limiting statistical precision

It cannot control for **publication bias** (studies with null or negative results may be underreported)

The **2% threshold is arbitrary** — while reasonable, it may not apply equally to all athletes or events

The conversion of time-to-exhaustion to time trial performance is an **approximation** with inherent error

The analysis is **descriptive rather than inferential** — no formal hypothesis testing across studies

**Major methodological weaknesses:**

**Non-randomized allocation:** Several studies allowed athletes to choose their diet based on belief, introducing powerful placebo effects and self-selection bias

**Order effects:** In crossover designs without adequate washout, LCHF was often undertaken after additional training, potentially improving fitness independently

**Suboptimal control conditions:** Many HCHO trials were conducted under fasted conditions or without carbohydrate during exercise, which is not representative of best practice sports nutrition

**Weight loss confounding:** Several LCHF groups lost significant body mass (2–6 kg), which independently improves performance on hills or weight-bearing events

**High dropout rates:** Some studies reported 7+ participants failing to complete the LCHF intervention, suggesting negative outcomes were systematically excluded

**Small sample sizes:** Individual studies ranged from n=1 to n=19, limiting statistical power and generalizability

**Primary finding: No evidence of performance benefit from ketogenic diets**

**No study showed a performance improvement exceeding the 2% threshold** that could be attributed to the LCHF diet itself, once confounds were accounted for

Studies that appeared to show positive effects (e.g., racewalkers in studies A and B) were **non-randomized** — athletes chose their diet based on belief, creating a strong placebo effect

In these non-randomized studies, the LCHF group showed ~1–3% improvement in 10,000-m race time, but the HCHO control group showed similar or greater improvements, or the comparison was confounded by weight loss

**Secondary findings: Detriment or no difference in well-controlled studies**

**Randomized crossover studies** (e.g., study F–J with 4–42 days of LCHF) showed **no significant difference** between LCHF and HCHO, or a trend toward worse performance with LCHF

**Study K** (4 weeks LCHF, crossover design): Larger individual variability in performance changes with LCHF than HCHO, with 2 additional subjects failing to complete the LCHF intervention — suggesting some athletes experienced substantial performance decrements

**Study L** (9 days LCHF, parallel-group): "Consistent reduction in endurance capacity" with LCHF compared to HCHO

**Study E** (4 weeks LCHF, crossover): Authors noted a "limitation of intensity of exercise that can be performed" with LCHF, describing a "throttling of function near V̇O2max"

**Findings from combined strategies (LCHF + carbohydrate restoration or ketone esters):**

**Study M** (3 weeks LCHF + 17 days HCHO periodization): No clear benefit over continuous HCHO; comparison was confounded by comparing different race distances

**Study N** (6 days LCHF + 1 day carbohydrate restoration): No performance benefit

**Study S** (7 days LCHF + ketone ester supplement): No performance benefit over HCHO

**Case study findings (Study O–R):**

One elite triathlete on 2 years of LCHF showed improved sprint power when adding exogenous carbohydrate, but this could not be separated from training effects or the acute carbohydrate benefit itself

**Summary across all 19 data points:**

**0 data points** showed a clear, confound-free performance benefit of LCHF over HCHO

**~5 data points** showed possible benefit but with major confounds (non-randomization, weight loss, order effects)

**~10 data points** showed no difference between LCHF and HCHO

**~4 data points** showed clear detriment with LCHF (reduced endurance capacity, inability to sustain high intensity)

To translate these findings into practical terms:

**No benefit detected:** Across all well-controlled studies, the difference between LCHF and HCHO was essentially zero — or negative. If there is any real effect of LCHF on endurance performance, it is smaller than the 2% threshold that matters in competition

**Weight loss confound:** In studies where LCHF groups lost 2–6 kg of body mass, any apparent performance improvement (e.g., on hilly courses) was likely due to the weight loss itself, not the diet composition. For a 70-kg athlete, losing 4 kg (~5.7% body mass) would improve running economy by roughly 1–2% on gradients — enough to appear beneficial but unrelated to ketosis

**Detriment in some athletes:** The "throttling" effect at high intensities means that for events requiring sustained efforts above ~80% of V̇O2max (e.g., 5-km to half-marathon races, criterium cycling), LCHF may reduce power output by 3–8%, which is a meaningful competitive disadvantage

**Individual variability:** Some athletes appeared to respond poorly to LCHF (dropping out of studies), while others maintained performance. This suggests that if there are responders, they are a minority, and there is no reliable way to identify them in advance

**What the authors acknowledge:**

They are "scratching the surface of the permutations of athlete, event, and LCHF strategy"

Greater transparency around individual data is needed to identify true responders vs. non-responders

There is a lack of investigations in true ultraendurance events (>6 hours), where LCHF might theoretically offer advantages

The 2% performance threshold is arbitrary and specific to the athlete and event

**What a critical reader would note:**

**Publication bias:** Studies with null or negative results for LCHF may be less likely to be published, yet even the published literature shows no benefit — the true effect may be worse

**Industry funding:** Several studies were conducted by researchers with a known advocacy position for low-carbohydrate diets (e.g., Noakes, Volek, Phinney), which may introduce unconscious bias in study design and interpretation

**Lack of female representation:** Only 2 studies included women, and neither controlled for menstrual phase, which affects substrate metabolism and performance

**Short adaptation periods:** Most studies used 3–4 weeks of LCHF, but full metabolic adaptation to ketosis (including upregulation of fat oxidation enzymes) may take 8–12 weeks. However, longer studies (e.g., 2-year case study) also showed no benefit

**No blinding:** It is nearly impossible to blind athletes to a ketogenic diet (extreme carbohydrate restriction is noticeable), so placebo effects and expectation bias are unavoidable

**Performance metrics:** Many studies used time-to-exhaustion protocols, which have higher variability and less real-world validity than time trials

**Dietary compliance:** Several studies used ad libitum diets with only education, meaning actual carbohydrate intake may have been higher than reported in the LCHF group

**No measure of ketone levels:** Not all studies confirmed that participants were in nutritional ketosis, so some "LCHF" groups may not have achieved the intended metabolic state

For someone running their own n=1 experiment to test whether a ketogenic diet improves their endurance performance:

### What to test

**Intervention:** A strict ketogenic diet (<50 g carbohydrate/day, ~75–80% fat, ~15–20% protein) for 4–8 weeks

**Comparator:** Your usual high-carbohydrate diet (or a moderate-carbohydrate diet with ~5–8 g/kg/day of carbohydrates)

**Dose:** The ketogenic diet must be strict enough to induce nutritional ketosis (blood ketones >0.5 mM, ideally 1–3 mM)

### Minimum meaningful duration

**At least 4 weeks** for metabolic adaptation (fat oxidation enzymes upregulate over 2–4 weeks)

**8 weeks is better** to allow full adaptation and to separate acute effects from chronic adaptation

Include a **2-week washout** between conditions if using a crossover design

### What to measure

**Primary outcome:**

**Time trial performance** on a set course or distance (e.g., 10-km run, 40-km cycling time trial) — not time to exhaustion, which is less reliable

Measure at baseline, after 4 weeks, and after 8 weeks on each diet

**Secondary outcomes:**

**Blood ketones** (beta-hydroxybutyrate) — confirm ketosis (target >0.5 mM)

**Body mass and body composition** (skinfolds or DEXA if available) — to separate diet effects from weight loss effects

**Rate of perceived exertion (RPE)** during standardized submaximal efforts

**Heart rate** during standardized efforts

**Sleep quality and recovery** (subjective scale)

**Mood and cognitive function** (to assess "keto flu" vs. adaptation)

### Key confounds to control for

**Training load:** Keep training volume and intensity identical between conditions (use a training log and match weeks)

**Weight loss:** If you lose weight on LCHF, this independently improves performance — you must account for this. Consider matching energy intake between conditions, or measure performance relative to body mass

**Hydration and electrolytes:** Ketogenic diets cause electrolyte shifts; ensure adequate sodium, potassium, and magnesium intake in both conditions

**Menstrual phase (if female):** Test at the same phase of your cycle in both conditions (e.g., follicular phase days 5–10)

**Time of day:** Test at the same time of day for all trials

**Pre-exercise nutrition:** Standardize the pre-exercise meal (or lack thereof) — but note that LCHF typically requires a low-carb pre-exercise meal, while HCHO allows carbohydrate loading

**Caffeine and supplements:** Keep consistent across conditions

**Placebo effect:** This is nearly impossible to blind, but you can reduce bias by having someone else randomize the order of conditions and not telling you which you're on (though you'll likely figure it out from the food)

### What a positive result would look like

A positive result for LCHF would be:

**≥2% improvement** in time trial performance on LCHF compared to HCHO (e.g., a 40-minute 10-km run becomes 39:12 or faster)

This improvement should be **consistent across multiple tests** (not just one lucky day)

The improvement should **not be explained by weight loss** — if you lost 3 kg, the performance gain could be from the weight loss alone, not the diet

You should also see **maintained or improved high-intensity performance** (e.g., sprint intervals, hill repeats) — if only low-intensity endurance improves but high-intensity suffers, the diet may not be suitable for your event

**A negative result** (which is more likely based on this meta-analysis) would be:

No change in performance (±1%), or

Performance decrement >2%, especially at high intensities

Inability to sustain training intensity (feeling "throttled" during hard efforts)

**Bottom line for your n=1:** The current evidence strongly suggests that for most endurance athletes, a ketogenic diet will not improve — and may harm — performance, especially in