Supplements · Diets
exogenous ketones increases blood ketones
In plain terms: Do ketone supplements actually raise blood ketones?
Strong support Supplements 💰 Industry COI noted
Part of: 🧪 exogenous ketones
RefutedContestedStrong support
consensus score 1.00
Yes — this is the one firmly proven effect, though the rise is only temporary.
📅 Last reviewed: 2026-07-15 ⓘ
Evidence ladder
How far up the ladder this claim has climbed. A high consensus on a low rung means "consistent so far," not "proven in people."
Top evidence so far: All trials, pooled (Meta-analysis)
MechanismIn-vitroAnimalObservationalRCTMeta-analysis
How the studies fall
18 support 0 contradict 0 tested null 0 mixed · 18 sources, 17 independent groups
What the evidence shows
Exogenous ketones (esters, salts, (R)-1,3-butanediol) reliably and dose-dependently raise blood BHB - the ONE unambiguously proven effect. Transient (~2-4h); monoesters give higher/faster peaks than salts or the 1,3-butanediol precursor (which needs hepatic conversion).
The evidence (21)
| Source | Grade | Stance | Quality | Finding |
|---|---|---|---|---|
| Stoner JT, Buga A, Crabtree CD, Robinson BT, Decker DD, Sapper TN, Kackley ML, Kraemer WJ, Volek JS 2026 · J Strength Cond Res | RCT | supports | moderate | Randomized, counterbalanced, placebo-controlled crossover, n=20 college-aged active adults (10 M / 10 F) over 2 visits separated by a 3-7 day washout, against an isocaloric/flavor-matched placebo: 50 g of the ketone precursor bis-octanoyl (R)-1,3-butanediol (BO-BD) raised capillary R-BHB, which 'rose incrementally post-BO-BD ingestion and peaked at 120-minutes (BO-BD vs. PL: 2.2 +/- 0.7 vs. 0.3 +/- 0.1 mM; p < 0.001)', sampled in 30-minute increments. Denominator is the full n=20 crossover sample - not a stratum, subset or completer subgroup. 2.2 mM is the PEAK of a still-rising curve at the last reported timepoint; the available abstract gives no values beyond 120 minutes, so this trial cannot speak to the ~2-4 h decay written into this claim's scope. Abstract-grade: blinding, registration, CI and funding/COI are all cannot-tell from the text available (the paper is not in the PMC open-access subset). A precursor requiring hepatic conversion, not a monoester. |
| Egan B 2025 · Scand J Med Sci Sports | observational | supports | low | [FT-verified, PMC12614059] SINGLE-AUTHOR NARRATIVE REVIEW, not a meta-analysis and not primary data. evidence_role SET TO review at reread 2026-08-25 per CONVENTIONS 5 and rubric axis 2: the paper's own data statement is 'Data sharing not applicable to this article as no datasets were generated or analyzed during the current study', and its stated purpose is to describe 'the rationale and potential mechanistic basis for the proposed effects of EKS on these outcomes, as well as critically appraising the existing literature in this field'. It pools nothing, has no protocol, no registration and no systematic search. The double count is literal and nameable, not theoretical: its reference [116] is Clarke, Russell, Spellanzon, Maher, Betts and Gonzalez 2025 (Eur J Nutr), which is s40794305 and already votes on THIS claim as a high-quality RCT - Egan re-describes that same trial's result here ('Blood [R-BHB] increased from ~1 mM at 30 min to ~4 mM at 4 h of recovery in KME'), so counting this row alongside it would count the Bath trial twice. STANCE UNCHANGED (supports): on the claim's own object the review is unambiguous - 'Ingestion of EKS produces an acute transient (1-2 h) increase in circulating AcAc and BHB concentrations', and 'Most studies that have focused on recovery and adaptation have used dosing strategies that produced [R-BHB] in a range of 2-5 mM'. DENOMINATOR: none exists. This is a narrative synthesis with no pooled n, no participant count and no effect estimate of its own; every concentration it cites belongs to a primary study, so the row carries a direction and no measurable magnitude. OFF-OBJECT and removed from this stance per rubric axis 1 - the previous extract's '(perf unproven)' is the exercise-performance object, on which the vault holds a separate claim (claim-exogenous-ketones-improve-exercise-performance) that this source does not appraise: 'the majority of studies have failed to observe a benefit [14 refs]. Negative effects on parameters of exercise performance and maximal exercise testing have also been observed in several studies [7 refs]'. GRADE FLAGGED NOT EDITED: 'observational' is wrong for a narrative review, but the closed enum has no review value and review-ness lives in evidence_role - same handling as the sibling row ev-s41097203 on this claim. Quality held at low (unchanged): single author, no systematic method, self-citing his own prior reviews [3, 7] as the authority for the performance verdict. Funding: 'The author received no specific funding for this work.' COI: 'The author declares no conflicts of interest.' SOURCE NOTE CORRECTED in the same edit - study_type read 'meta-analysis' for a paper whose title ends 'A Narrative Review', which is exactly the mislabel that let the 2026-08-16 zero-weight pass miss it. |
| Marcotte-Chénard A et al 2026 · Am J Physiol Cell Physiol | observational | supports | high | Single-arm, unblinded ACUTE PRE-POST study - not randomised, no control or placebo arm in the available text; every contrast is 'Compared with baseline'. 13 healthy adults (7 F / 6 M, age 28 +/- 7 yr) consumed 0.750 g/kg body mass ketone monoester [(R)-3-hydroxybutyl (R)-3-hydroxybutyrate], blood sampled before and 2 h after: 'Blood BHB concentration significantly increased after KME consumption (P < 0.0001), peaking at 2 h (5.0 +/- 0.8 mmol/L)' - the highest peak among this claim's rows, matching the large 0.750 g/kg dose. Grade moved RCT -> observational at reread 2026-08-25 under the exposure-randomised test (CONVENTIONS S5 addendum 3); same shape as sibling row ev-s41245884 on this claim. BHB was the manipulation check, not the paper's primary endpoint: the primaries were monocyte lysine Kbhb (~70% rise, P = 0.004) and plasma BHB-amino-acid conjugates (P < 0.0001), both recorded as candidates. Abstract-grade read - Am J Physiol Cell Physiol 2026;330(3):C643-C650, subscription-only, no PMCID; funding/COI not seen. Independence flag: senior author Little JP also authors ketone-bhb-ma-2022 and s36847287, two further counting rows on this claim, under different group_ids. |
| Falkenhain K, Daraei A, Little JP 2024 · J Diet Suppl | RCT | supports | moderate | Double-blind randomised crossover PILOT, N=12 healthy adults (29 +/- 5 y, BMI 25 +/- 4 kg/m2, 42% female); three experimental trials each providing 10 g of active ingredient - (i) the monoester (R)-3-hydroxybutyl (R)-3-hydroxybutyrate, (ii) D-beta-hydroxybutyric acid with R-1,3-butanediol, (iii) R-1,3-butanediol - with capillary beta-OHB and glucose sampled at baseline and across 240 min. 'beta-OHB was elevated compared to baseline in all conditions'; total and incremental AUC (p < 0.05) and peak beta-OHB (p < 0.001) differed between conditions, 'with highest values seen in the ketone monoester condition'. WHAT THE DESIGN CAN AND CANNOT CARRY: all three arms received an exogenous ketone supplement and there was no placebo condition, so the randomisation bears on WHICH supplement (a genuine randomised between-supplement comparison) while the ketones-raise-BHB finding that this claim rests on is a within-subject pre-post comparison - grade FLAGGED at reread 2026-08-25 as arguably observational for this claim per CONVENTIONS 5 addendum 3, not edited. GLUCOSE CORRECTION: the previous extract read 'monoester highest, also lowered glucose', implying a monoester-specific glucose effect the paper denies - 'Blood glucose was reduced after consumption of each supplement, with no differences in total and incremental area under the curve across supplements.' Abstract-grade and permanently so (Europe PMC: no OA full text, subscription-only), so NO BHB or glucose concentrations exist for this source - p-values only. Funding/COI not seen. Same group (Falkenhain/Daraei/Little, UBC Okanagan) as ketone-bhb-ma-2022 on this claim under a different group_id - flagged, not fixed. |
| Seto, Nardone, Thiessen, Teixeira, Bommarito, Pocchi, McCarthy, Millar 2025 · Physiol Rep | RCT | supports | high | [FT-verified] Double-blind randomized crossover, n=19 healthy young adults, NCT04881526: oral ketone monoester 0.4 g/kg raised capillary beta-HB to 0.9+/-0.5 mM at 20 min and 1.5+/-0.8 mM at 40 min vs 0.2+/-0.1 and 0.3+/-0.2 mM on isocaloric MCT placebo (p=0.0001 and p=0.00003); baseline identical (p>0.99). Magnitude caveat the authors themselves raise: mean peak ~1.5 mM and no participant exceeded 3 mmol/L, below the 3-4 mM reached in the studies reporting cardiovascular effects. Capillary handheld meter, not plasma assay. |
| Clarke L, Russell SL, Spellanzon B, Maher JL, Betts JA, Gonzalez JT 2025 · Eur J Nutr | RCT | supports | high | Single-blind randomised crossover (NCT06846840, University of Bath, >=7 d washout, median 14 d), 18 recruited / 13 completed endurance-trained MEN ONLY (29 +/- 12 y, range 18-61; VO2peak 62 +/- 8 mL/kg/min). Both conditions ingested sucrose 1.0 g/kg/h + whey 0.4 g/kg/h every 30 min through a 4 h post-exercise recovery; the ONLY randomised difference was ketone monoester (R)-3-hydroxybutyl (R)-3-hydroxybutyrate at 0.29 g/kg/h versus an isoenergetic taste-matched olive-oil placebo, so the claim's own exposure is the randomised variable. ON THIS CLAIM: blood BHB 'remained below the limit of detection during PLACEBO, but rose to ~ 4 mmol L-1 during KETONE and remained elevated compared with PLACEBO through recovery and the second running bout' (condition x time interaction p < 0.0001); the abstract states the same contrast as > 3.5 mmol L-1 versus PLACEBO, p < 0.0001. DENOMINATORS DIFFER WITHIN THE PAPER: n = 13 carries the whole 4 h recovery BHB curve, but n = 12 carries every run-2 datum including the persistence of ketosis into the second run, one participant having been injured during his PLACEBO time-to-exhaustion test. Measured by hand-held point-of-care meter (Optium Neo), not a lab assay, and 5% of all data points (103 of 2003) were mean-imputed from flanking timepoints. Quality reads high rather than moderate on what the paper established about itself: registered, placebo-controlled, a full time course with an interaction test rather than a single manipulation-check draw, and a blinding exit interview in which 62% noticed a drink difference but only 38% identified their condition correctly, i.e. no better than chance. Funded by the University of Bath with no external funding and no ketone-industry sponsor; senior authors JTG and JAB declare extensive food-and-beverage-sector ties (ZOE, PepsiCo, Lucozade Ribena Suntory, Metabolic Solutions Ltd) but none to ketone esters. Registered placebo was medium-chain triglyceride, changed to olive oil before enrolment and disclosed, because MCT can itself induce ketosis. OFF-OBJECT AND DELIBERATELY LEFT OUT OF THIS STANCE per rubric axis 1 - the previous extract's clause 'lowered post-exercise glucose' belongs to claim-exogenous-ketones-decrease-postprandial-glucose, on which this source carries NO appraisal, so it must not ride inside the BHB vote: glucose fell by > 1.2 mmol/L by the end of recovery (p = 0.007) with insulin unchanged (p = 0.89), and the difference was gone by 15 min into run 2 (p = 0.64) and at exhaustion (p = 0.57). Also off-object: no difference in time-to-exhaustion (54 +/- 33 vs 52 +/- 28 min, p = 0.87, n = 12), increased sucrose retention (206 +/- 26 to 220 +/- 26 g, p = 0.001), and lower bicarbonate with unchanged pH. EXTRACT CORRECTED at reread 2026-08-25: the previous text carried no number, no dose, no comparator and no denominator on a high-quality-RCT weighting. |
| Kjaer 2026 · JMIR Diabetes | RCT | supports | low | [FT-verified, PMC12799078] Research Letter reporting the continuous-ketone-monitor (CKM) sub-study of a randomised ketone-ester trial (NCT06053138, Aarhus); the parent study on exogenous ketosis and erythropoiesis is stated as unpublished. DENOMINATOR IS THE SUB-SAMPLE, NOT THE TRIAL, and the previous extract named none: 'CKM became available midway through the study and was therefore applied sequentially in the final 7 of the 16 healthy volunteers' - 4 KE (3 female, 1 male) and 3 PBO (2 female, 1 male), median age 41 y (IQR 28-55), with one KE participant's sensor detaching on day 4 and not replaced. The claim's own exposure IS the randomised variable: 'Participants were randomized to receive either a ketone ester (KE) drink (500 mg/kg/d) or a placebo (PBO), matched for volume, taste, and viscosity', taken two to three times daily for two weeks. ON THIS CLAIM: 'BHB AUCs were significantly influenced by both day and treatment, with an interaction effect (P=.006)', and the discussion is explicit that 'the CKM detected increases in interstitial BHB concentrations following KE ingestion', the arms separating until they converged on the final day. NO CONCENTRATIONS EXIST: not one mmol/L value appears anywhere in the letter - BHB enters only as log-transformed daily AUC, shown as back-transformed least-square means in Figure 1 - so this row votes a direction with no magnitude, measured in interstitial fluid by an unvalidated device (SiBio KS1), not in blood. STANCE CORRECTED mixed -> supports at reread 2026-08-25 per rubric axis 1: the disconfirming half is OFF-OBJECT. The 14-day decline also occurred under placebo - 'a smaller but significant decline was also observed in the PBO group (P=.02)' - and the authors assign it to the instrument, 'likely reflecting sensor drift rather than physiological adaptation', so it is a finding about the monitor and not about whether exogenous ketones raise BHB; it is recorded as a candidate instead. Quality held at low: a post-hoc sub-sample of 7, 'investigators were not blinded', 'the study was not originally designed to evaluate CKM performance', and the authors' own decisive limitation - 'not comparing the CKM-derived ketone levels with gold standard blood BHB measurements (eg, finger-prick tests), making it difficult to definitively decide if our observations are due to sensor-specific limitations or not'. DUPLICATE-POPULATION RISK, flagged not applied: if the parent trial NCT06053138 is ever ingested, these same 7 participants would vote twice. Funding: internal Regional Hospital Central Jutland research funds; 'Conflicts of Interest: None declared.' |
| Lokken 2022 · J Inherit Metab Dis | RCT | supports | high | Randomised, double-blind, placebo-controlled crossover (NCT03945370, visits >=1 week apart): 395 mg/kg oral D-BHB monoester vs taste-matched placebo 25 min pre-exercise, n=12 total - 8 genetically confirmed McArdle/GSDV patients and 4 healthy controls, each group carrying its own denominator below. In the 8 GSDV patients plasma HOB rose from a fasted 78.6 +/- 55.8 to 2378.6 +/- 1248.0 umol/L at 15 min and 3289.9 +/- 1327.0 umol/L at exercise onset, where it plateaued (p = 0.00008; AcAc p = 0.0001), and KB oxidation rose 0.6 +/- 0.5 to 4.3 +/- 1.3 umol/kg/min (p = 0.0001); the 4 controls showed the same rise (oxidation 1.1 +/- 0.4 to 5.3 +/- 0.7, p = 0.00006), with no between-group difference in the ketone response. The glucose fall is the GSDV subgroup's own result (n=8, p = 0.020, 5.75 predrink to 4.35 mmol/L end-exercise); in the 4 controls it was only a trend (p = 0.087), and tracer-measured glucose Ra did not differ in either group (p = 0.628 / 0.265) despite the abstract saying it was lowered. BHB was a secondary outcome - the trial was powered on heart rate, which did not improve (GSDV 117.9 +/- 14.3 placebo vs 123.0 +/- 15.2 KE, p = 0.120). Independent of the ester industry: Lundbeck Foundation funded, investigator driven, no author COI. |
| Miyatsu T, Tate C, McAdam J, Massey C, Broderick T 2025 · Metabolism Open | observational | supports | moderate | [FT-verified, PMC12617643] Single-group observational pre-post, NO control arm, no randomisation, no placebo: a 4-day unsupplemented baseline phase then 10 days of twice-daily weight-stratified ketone monoester (DeltaG Tactical Ketones, TdeltaS Global), approx 415-442 mg/kg post-breakfast (27 / 35 / 42 g by weight band) plus a one-third booster 1.5 h later, in 20 healthy adults aged 18-35 (10 M / 10 F). DENOMINATOR CORRECTED at reread 2026-08-25: the pharmacokinetic numbers this row votes on are NOT on n = 20 - 'In total, this data exclusion process resulted in removal of 4 participants and 10 % of total monitoring days from the remaining 16 participants', and the authors' own statistics section says 'Given the modest sample size (n = 16)'. The n = 20 denominator carries only the blood-biomarker and sleep analyses. NUMBERS ADDED, the previous extract carrying none on a claim about how much BHB rises: Cmax 5.27 +/- 0.93 mmol/L against a stated physiological fasting range of 0.2-0.5 mmol/L, Tmax 1.17 +/- 0.63 h, onset to 0.5 mmol/L 0.24 +/- 0.23 h, offset 5.12 +/- 1.31 h, AUC 12.11 +/- 4.69 hmmol/L, half-life 1.97 +/- 0.76 h. ENDPOINT AND MODALITY: interstitial-fluid BHB by Abbott Keto Lingo continuous ketone monitor (cited MARD ~14 % and r > .90 versus capillary BHB), not a plasma assay - and the device truncates the very endpoint, 'the CKM device had an upper detection limit of approximately 6 mM ... and several participants occasionally reached or briefly exceeded this threshold', so Cmax and AUC are floors rather than point estimates. The ~5 h offset is a property of THIS protocol, not of the dose: the authors state the booster 'extended the observed ketone elevation, influencing both the offset time and total ketone exposure', and the single-dose half-life is 1.97 h. Quality held at moderate (unchanged): prospective, dosing adherence confirmed by daily video-call proctoring, sensor site and insertion standardised, limitations enumerated by the authors - against an uncontrolled within-subject contrast, an exclusion threshold 'determined post-hoc after visual inspection of data completeness', and a fifth of participants dropped. Not low, because the ceiling error runs toward the claim and the effect dwarfs the named noise; not high, because nothing was controlled. Funding: 'The study was an IRB approved, internally funded study' (IHMC IRB-2022-0040, no trial registration). COI: 'Abbott Lingo provided continuous ketone and glucose monitoring devices as in-kind support'; no declared relationship with TdeltaS Global, the monoester's maker. OFF-OBJECT and deliberately kept out of this stance per rubric axis 1, recorded as candidates instead: acute glucose suppression (22.09 +/- 10.43 mg/dL, Granger causality significant at 5 min and 25-55 min post-dose, belonging to claim-exogenous-ketones-decrease-postprandial-glucose on which this source carries no row), the fasting-glucose RISE after 10 days (p = .04), and reductions in sleep efficiency, duration and score. |
| Roca GM, Barker GA, Waldman HS 2026 · J Strength Cond Res | RCT | supports | moderate | [abstract-grade, PubMed efetch 2026-08-25; J Strength Cond Res 2026 Jul 10 online ahead of print, doi 10.1519/JSC.0000000000005606 - Europe PMC gives pmcid null, isOpenAccess N, inEPMC N and a single 'Subscription required' route, so no full text exists for this source and it is abstract-grade permanently] Double-blind, randomised, placebo-controlled crossover: 25 healthy adults (20 men, 5 women) 'completed 3 trials (KME + CAF, KME, placebo [PLA])' around a 20-minute mental-fatigue protocol. On this claim's object: 'Both KME conditions elevated βHB to ∼1.5 mM, confirming nutritional ketosis.' DENOMINATOR: 25 is the full crossover completer sample - not a stratum, subset or per-outcome subgroup - and each participant contributes all three conditions. WHAT THE TEXT CANNOT CARRY, corrected at reread 2026-08-25: the previous extract read 'vs placebo', but the abstract states no PLA βHB value, no dispersion, no p-value and no confidence interval for that contrast - the elevation is reported, the comparison is not quantified. The KME DOSE is not stated and no time course or peak timing is given, so this row speaks only to the fact of an acute rise and bears on neither the dose-dependence nor the ~2-4 h transience written into this claim's scope. βHB was a manipulation check; the primary endpoints were task-switching and Stroop reaction time, recorded as candidates along with an explicit null on subjective fatigue and workload. Grade RCT HELD under the exposure-randomised test (CONVENTIONS 5 addendum 3): unlike siblings ev-s41604251 (single-arm pre-post) and ev-s36847287 (no placebo condition), the claim's own exposure was randomised against a placebo arm here. Quality held at moderate: strong design, but the βHB finding carries no numbers beyond the ~1.5 mM point and no registry, funding or COI is visible - limits of the available text, not established defects. ~1.5 mM matches the monoester peak in ev-s41047637 and sits far below the 5.0 mM reached on 0.750 g/kg in ev-s41604251. Independence: Roca GM / Barker GA / Waldman HS (North Alabama, North Georgia) appear on no other source note in the vault. |
| Holland-Winkler AM, Moore AR, Bederman I 2025 · Nutrients | RCT | supports | moderate | Randomized, triple-blinded, placebo-controlled crossover PILOT in healthy untrained young adults, 13 analysed of 16 enrolled (21.6 +/- 4.3 y, 7 M / 6 F, BMI 25.7); one 7 g dose of racemic sodium D,L-betaHB (50% D / 50% L) vs flavour- and calorie-matched maltodextrin placebo, 10 h fasted, single blood draw 30 min post-dose, one-week washout. THE MEASURED ENDPOINT IS ACETOACETATE, NOT betaHB: plasma AcAc by GC/MS rose PRE 1.61 +/- 0.12 to POST 2.18 +/- 0.41 mM (p < 0.001, d = 1.28) with no change on placebo (1.59 -> 1.66, p = 0.307), gain-score contrast 0.57 +/- 0.44 vs 0.07 +/- 0.23 mM (p = 0.009, d = 0.86); urinary ketones 0.00 -> 27.91 +/- 29.05 mg/dL on a 12-subject denominator (one further subject's urine data removed). No betaHB concentration is reported anywhere in the paper - betaHB enters only as the correlate of DeltaAcAc (r = 0.757, p < 0.001, KS condition), so this source carries a blood-ketone rise, not a betaHB number. Authors' own limits kept: 'we cannot verify whether the increase in AcAc that was observed was due to the betaHB that was ingested'; 2 of 13 did not respond (one no increase, one a decrease); and their peak AcAc is ~10x the only comparable racemic-salt study (2.18 vs 0.23 mM) at roughly HALF the dose, a gap they attribute to other methods underestimating without testing it - baseline alone (1.61 mM, fasted) exceeds that study's peak. No registration identifier; no external funding, no declared COI; power stated >0.80 for primary findings. EXTRACT CORRECTED at reread 2026-08-25: the previous text's '(modest vs ester)' asserted a comparison this paper never ran - there is no ester arm, and the authors conclude the opposite, that conversion occurs 'just as with ketone ester supplements'. |
| Panse N, Lafrenaye A, Bergman C, Gerk PM 2025 · Int J Pharm | animal | supports | moderate | Rat pharmacokinetic study (Sprague-Dawley rats) of Veech ketone monoester (KE), the main ingredient of a popular exogenous ketone supplement, given by IV bolus, IV continuous infusion and orally; plasma KE and its immediate metabolites BHB, 1,3-BD and AcAc were tracked as concentration-time profiles and used to fit PK parameters. ON THIS CLAIM the abstract's only directional statement is that administration 'lowered the blood glucose level in a manner inversely mirroring the elevation of plasma BHB levels' by BOTH the oral and IV routes - i.e. BHB rose - with BHB's half-life about 13 min (KE itself 2-3 min, AcAc 23 min, 1,3-BD 26 min). DENOMINATOR AND ENDPOINT, stated as cannot-tell rather than as absence: no animal count, no dose, and no BHB Cmax or AUC are reported in the abstract, and whether a vehicle-control arm existed cannot be determined from it, so this row votes on a qualitative direction without a number attached - it is a real observation, not a measured effect size. The paper is Elsevier subscription-gated with no PMC deposit; NCBI efetch and Europe PMC were both probed at reread 2026-08-25 and neither yields full text. Oral absolute bioavailability of the INTACT ester was 'extremely low at about 3%', which is a fact about KE the molecule and not about BHB delivery - the ester is hydrolysed to BHB and 1,3-BD - so it does not cut against the oral BHB rise reported in the same paper. OFF-OBJECT and deliberately excluded from this stance per rubric axis 1: the glucose fall that the previous extract carried inside this BHB vote is a different object and is recorded as a candidate instead (it cannot be filed on claim-exogenous-ketones-decrease-postprandial-glucose, whose object is postprandial glucose, because the rats' feeding state is not stated); likewise the 24 h continuous-infusion plasma chemistry - no significant change in plasma pH, significant reduction in blood glucose and blood urea nitrogen. EXTRACT CORRECTED at reread 2026-08-25: stance (supports), grade (animal) and quality (moderate) all reconfirmed and unchanged - quality held at moderate because an abstract that does not mention a control has not established that there was none. |
| Mohib O et al. 2025 · Nutrients | observational | supports | moderate | [FT-verified, PMC12525594] PRISMA systematic review (PROSPERO CRD42023492846) of 51 clinical studies of exogenous ketosis in adults with disease. IT POOLS NOTHING OF ITS OWN: 'Due to the heterogeneity of study populations, interventions, and outcomes, a quantitative synthesis was not feasible. Instead, we conducted a structured narrative synthesis.' evidence_role SET TO review at reread 2026-08-25 per CONVENTIONS 5 (a narrative/systematic review reports no new data and would double-count primaries the vault already holds - Nielsen 2019, one of the trials it summarises, votes here separately as ketone-hf-2019). The 2026-08-16 zero-weight pass caught this paper's cardiac-output row and missed this one because the source note's study_type read 'meta-analysis'; that field is corrected in the same edit. SUBSET DENOMINATORS THE PREVIOUS EXTRACT ERASED with the words 'supplements consistently elevate blood BHB': of the 51 studies, 25 used ketone esters, 5 ketone salts and 21 MCTs, and on that 21-study MCT fifth the review's own verdict is the opposite of consistent elevation - 'Most clinical trials utilized MCTs, achieving relatively low BHB levels (often <0.5 mmol/L), which may be insufficient to elicit robust clinical results', i.e. below the paper's own threshold, 'Ketosis is defined as a blood BHB level greater than 0.5 mmol/L'. Its abstract says the same: 'Subtherapeutic ketone concentrations induced by medium-chain triglycerides ... often precluded firm conclusions'. One 4-month multiple-sclerosis MCT trial (n=51) reports 'Tested after an overnight fast. Median BHB: 0.05 (range 0.33) mmol/L'. The authors also state 'MCTs are not exogenous ketone supplements per se', so the MCT fifth is only partly on this claim's subject. STANCE UNCHANGED (supports) because the ester and salt studies tabulated here do raise BHB into the 1-4 mmol/L range - what changes is that the row no longer votes. Funding: FWO G075423N and Wetenschappelijk Fonds Willy Gepts; 'No conflicts of interest are declared for the authors of this manuscript.' Grade left at observational and FLAGGED rather than edited - a narrative SR is neither, and the same paper is graded RCT on its other row. |
| Forester G, Swindle S, Shott ME, Frank GKW 2026 · Int J Eat Disord | RCT | supports | moderate | Randomized, double-blind, placebo-controlled CROSSOVER, n=22 adults across the anorexia/bulimia spectrum (AN n=8; OSFED-AN n=9; BN n=4; OSFED-BN n=1), ages 19.0-52.1 y, BMI 17.2-46.2 - a clinical, wide-BMI eating-disorder sample, not healthy volunteers. 'On separate days, participants consumed 10 g beta-hydroxybutyrate (BHB) or placebo prior to completing a taste-based conditioning task during electroencephalography (EEG).' The claim's own exposure IS the randomised variable, so RCT stands (rubric axis 4). ON THIS CLAIM THE PAPER REPORTS EXACTLY ONE SENTENCE: 'Active BHB increased blood ketone levels.' EXTRACT CORRECTED at reread 2026-08-25 (rubric axis 6): the previous text read '10g BHB dose confirmed to increase blood ketone levels vs placebo', which presents a measured contrast that does not exist - the abstract gives NO concentration, no units, no timepoint, no p-value and no assay method for blood ketones, and states no vs-placebo test for them (the 'vs placebo' is design-implied, not reported). DENOMINATOR: n=22 is the study's headline sample for an EEG endpoint; whether blood ketones were assayed in all 22 participants, and in both sessions, is cannot-tell from the abstract, so n=22 must not be read as the ketone measurement's own denominator. Blood ketones here are a manipulation check for a proof-of-concept EEG study whose real endpoint is an event-related potential. CANNOT SPEAK TO THIS CLAIM'S SCOPE ('acute, dose-dependent, transient ~2h; esters>salts>1,3-butanediol'): the BHB FORMULATION is unspecified (salt, free acid, ester or precursor - the abstract says only '10 g beta-hydroxybutyrate'), only one dose was tested, and no time course is reported. Quality HELD at moderate: the randomised, double-blind, placebo-controlled crossover is a genuine reliability statement, while nothing was established about the reliability of the ketone measurement itself - and an abstract's silence is not a defect to infer. Abstract-grade (NCBI efetch; Int J Eat Disord, online ahead of print). Funding and COI not seen; no registry stated. OFF-OBJECT and deliberately left out of this stance per rubric axis 1 - the paper's actual finding, 'Compared to placebo, BHB was associated with a significant reduction in LPP sensitivity to unsigned prediction error (p = 0.028)', belongs to a neural-signalling object the vault holds no claim for; it is recorded as a candidate. INDEPENDENCE FLAGGED NOT FIXED: senior author Frank GKW and co-author Shott ME (UCSD) also carry s42236862, which votes on claim-keto-reduces-depression-anxiety under a different group_id - no double count on THIS claim, but the source note's truncated 'Forester G et al' hid the overlap entirely. |
| Valenzuela 2024 · Clin Nutr | RCT | supports | moderate | Randomized cross-over trial (blinding not stated - cannot-tell at abstract grade), n=15 in two groups each carrying its own denominator: 8 McArdle (GSDV) patients and 7 healthy controls, comparing placebo against 30 g of a D-beta-hydroxybutyrate / D-1,3-butanediol monoester before a submaximal constant-load test followed by a maximal ramp test; patients were additionally assessed after 75 g carbohydrate. On THIS claim the paper reports one line: 'Ketone supplementation induced ketosis in all participants (blood [ketones] = 3.7 +/- 0.9 mM) and modified some gas-exchange responses (notably increasing respiratory exchange ratio, especially in patients).' That is total blood ketones at a single descriptive mean with no placebo-comparison statistic, no time course and no BHB-specific value - a manipulation check inside a trial whose stated aim and every reported endpoint are exercise capacity, which is why quality reads moderate rather than high despite the randomised design. EXTRACT CORRECTED at reread 2026-08-25: the previous text ('markedly raised plasma ketone bodies and KB oxidation ... lowered glucose/FFA') was a paraphrase of a DIFFERENT trial, Lokken 2022 (s35150142), which sits on this same claim - Valenzuela 2024 reports no KB-oxidation, no glucose and no free-fatty-acid result anywhere. Confirmed independent of Lokken: different site and investigators (lucia-madrid vs vissing-copenhagen, zero author overlap), different patients (Spanish vs Danish, 8+7 vs 8+4), different dose (30 g fixed vs 395 mg/kg), different primary endpoint (peak power output vs heart rate); Lokken registers NCT03945370 while no registration identifier appears for this trial. Abstract-grade permanently (Clin Nutr, isOpenAccess=N, inEPMC=N). Funder per Europe PMC: Instituto de Salud Carlos III; COI not seen. OFF-OBJECT and left out of the stance per rubric axis 1: ketones IMPAIRED peak power in the 8 patients (-11.6 % vs placebo, p = 0.001) with no effect in the 7 controls (p = 0.268). |
| Falkenhain K, Daraei A, Forbes SC, Little JP] # full byline read at reread 2026-08-25 2022 · Adv Nutr | meta-analysis | supports | moderate | Between-condition pool of 33 comparisons (NOT the review's 43 trials / 586 participants): blood BHB MD = +1.98 mM (95% CI 1.52, 2.45; P < 0.001) but I2 = 98.2%; monoesters +2.65 mM vs salts +0.48 mM (P < 0.001) and monoester dose moderates BHB (P < 0.001). Within-group pool, also 33 comparisons, MD = +1.73 mM (95% CI 1.26, 2.21), I2 = 99%. Direction is unanimous across every subgroup; magnitude is not, and the authors state that outside the between-condition glucose analysis they saw 'a large degree of heterogeneity alongside wide prediction intervals... which could indicate that certain populations may not experience the observed' effect. 37 of 43 included studies at 'some concern' for risk of bias (6 low, none high), with acknowledged ketone-industry conflicts among included trials. No publication-bias test is reported for the BHB analyses - cannot-tell. |
| Graybeal AJ et al 2025 · Physiol Rep | RCT | supports | moderate | [FT-verified, PMC12661118] Randomised single-blind placebo-controlled matched-pairs CROSSOVER (order randomised via random.org, >=4 d washout, no order effects all p >= 0.075), prospectively registered NCT05651243; 282 mg/kg body mass ketone monoester (DeltaG Tactical, TdeltaS Global) in 250 mL water versus a volume- and taste-matched Bitrex/stevia/cellulose placebo, so the claim's own exposure IS the randomised variable. DENOMINATOR CORRECTED at reread 2026-08-25: n = 16 is not the enrolled cohort - 10 per group were recruited and 'two participants from each group ... had incomplete respiratory gas data across multiple timepoints due to device malfunction. Thus, these participants were excluded from this evaluation', leaving 8 MetS (5 M, 3 F) and 8 matched non-MetS. ON THIS CLAIM: condition x time p < 0.001 (eta2p = 0.94) for BHB, and 'During the KE trial, BHB increased across all time points in the combined, MetS, and non-MetS groups (Figure 2a,b; all p < 0.001)'. SUBSET NAMED, correcting the previous extract's unqualified parenthetical that the MetS group showed a higher BHB response: the between-group difference was at ONE timepoint - 'BHB was significantly higher for the MetS than the non-MetS group at 105 min post ingestion (p = 0.030; d: 1.21; d 95%CI: 0.12, 2.27)' - not across the curve, and the authors read it as reduced ketone clearance in MetS rather than greater delivery. NO CONCENTRATIONS ARE REPORTED: no Cmax, Tmax, AUC or any mmol/L value appears in the text; BHB curves exist only in Figure 2a,b, so this row votes a direction with no magnitude, measured in duplicate by point-of-care meter (Keto-Mojo GK+) at three timepoints (baseline, 45, 105 min) after an >=8 h overnight fast. PLACEBO WAS NOT INERT, which cuts slightly against the contrast: 'For the combined and non-MetS groups, BHB significantly increased during the placebo condition beginning at 45 min post-ingestion (combined: p = 0.021; non-MetS: p < 0.001), but only one participant reached a BHB concentration >=0.50 mmol/L (105 min post-ingestion) indicating the absence of ketosis at the group level' - a fasting drift, not a treatment effect. Quality held at moderate: registered, placebo-controlled, taste-matched, baseline BHB not different between trials (all p >= 0.196), 83.7% power for the interaction - against n = 16, 'investigators were not blinded to treatment conditions', a point-of-care rather than plasma assay, and a figure-only endpoint. DUPLICATE-POPULATION RISK, flagged not applied: this is an explicit secondary analysis and 'Only a limited set of secondary variables, including descriptive characteristics and capillary concentrations of beta-hydroxybutyrate (BHB) and blood glucose, are shared across related investigations (Graybeal et al., 2025; Stavres et al., 2025)' - the BHB data this row votes on are the shared ones. Neither parent is in the vault today, so the row is not zeroed; if Graybeal 2025 or Stavres 2025 (both NCT05651243) is ever ingested, this row must become duplicate-population. Funded by NIGMS U54GM115428; the authors declare no conflicts of interest. |
| Lyksholm TZ, Duus CL, Nielsen SF, Thomsen HH, Bech JN 2026 · Physiol Rep | RCT | supports | high | Double-blind, placebo-controlled randomized crossover (13 completers of 15 randomized, healthy adults 18-30): ketone monoester 300 mg/kg three times daily for 5 days raised plasma BHB vs matched placebo (p<0.001), with peaks of 3.0 mmol/L (95% CI 2.0-4.4) and 3.7 mmol/L (95% CI 2.4-5.6) about 30 min after each dose and morning pre-dose BHB still 1.97-fold above placebo (95% CI 1.11-3.46, p=0.020). BHB was the pharmacodynamic measure; the trial's primary endpoint was measured GFR. |
| Siddiqi AK et al 2026 · World J Crit Care Med | meta-analysis | tested-null | low | [FT-verified, PMC13248031] OFF-SCOPE on the object: this meta-analysis reports no blood-ketone measurement. Its extracted outcomes are 'CO, systemic vascular resistance, LVEF, heart rate, venous oxygen saturation, pulmonary capillary wedge pressure, and other cardiac functional indices' (Methods adds LVEDV, LVESV, GLS, S'max and TAPSE) - eleven haemodynamic and echocardiographic endpoints, no circulating BHB among them, and no ketone concentration appears anywhere in the results. That exogenous ketones raise circulating ketone levels is the paper's PREMISE ('Ketone bodies, especially beta-hydroxybutyrate, have shown the potential to improve hemodynamic outcomes'), never its measurement. The previous extract - 'BHB infusion/ingestion raised circulating ketones and improved hemodynamics' - asserted a ketone-concentration result the paper does not contain. Zero weight from here; the paper's real vote is on cardiac output, where it counts. Stance left as recorded rather than flipped, because the row is zero-weight once the role is set and no stance can be earned on an endpoint the paper never assessed. |
| Prins PJ et al 2026 · J Am Nutr Assoc | RCT | supports | moderate | Single-blind RCT, 18 recreational runners randomised 9 KME vs 9 isocaloric flavour-matched placebo, 30 g/dose thrice daily (90 g/day) for 31 days. In the KME arm (n=9) post-ingestion R-BHB was reliably elevated to about 3.0 mM (p<0.001), so thrice-daily chronic dosing did not blunt the acute rise. The 15% lower glycemia (p<0.001) is the same n=9 arm and is off-object here - it belongs to the postprandial-glucose claim, and the abstract does not say which glycemic measure fell. Abstract-grade read. |
| Thiessen J et al 2026 · Appl Physiol Nutr Metab | RCT | supports | moderate | [Abstract-grade, NCBI efetch; Appl Physiol Nutr Metab 2026;51:1-6, no PMC full text] Randomized double-blind placebo-controlled CROSSOVER in 18 healthy adults (23 +/- 3 y, 10 female) ingesting 0.3 or 0.6 g/kg ketone monoester or placebo. The claim's exposure IS the randomised variable, so RCT stands. ON THIS CLAIM the abstract says only: 'Compared to placebo, both ketone doses increased plasma beta-hydroxybutyrate'. ENDPOINT/DENOMINATOR CORRECTED at reread 2026-08-25 (rubric axis 6): the previous extract read 'both significantly increased plasma beta-hydroxybutyrate vs placebo (dose-dependent)', and neither qualifier is the paper's. NO p-value, NO concentration and NO dose contrast is reported for BHB - the two p-values ('at 45 min (both p < 0.02) and 120 min post-ingestion (both p < 0.01)') attach to the pair pRR50 and RMSSD, and the title's 'in a dose-dependent manner' qualifies cardiac vagal modulation, not the ketone rise. So this row votes a direction with no magnitude: BHB is the manipulation check for a cardiac-autonomic study whose primary finding is 'Acute ketone ingestion decreases HR variability measures associated with cardiac vagal modulation' (recorded as a candidate; the vault holds no claim on ketones and autonomic modulation). Quality held at moderate - sound blinded crossover but n=18 healthy 23-year-olds, acute single dose, and abstract-only with no BHB numbers to check; same label as the sibling abstract-grade crossovers ev-s41773904 and ev-s42294671. INDEPENDENCE FLAGGED, NOT FIXED: same laboratory as the sibling counting row s41047637 (shared authors Thiessen J, Pocchi G, Millar P; same 'Human Cardiovascular Physiology Laboratory', University of Guelph), which is grouped 'millar-guelph' while this source carries the default group_id - one group counting as two voices; participant overlap cannot be told from the abstracts and the designs differ. COI: 'The authors have no competing interests to declare'; funding not stated. |
Disagree, or know a study we missed?
We grade by evidence, not opinions. The way to weigh in is to point us to a study we haven't cited (check the evidence table above first), or to flag a problem with one we have. Every submission is reviewed; if it holds up, the grade updates and shows in Science Changes Its Mind.
Opens a short form. You'll sign in with Google so submissions are tied to a real account — we don't display your identity, and we only accept a link we can verify (PubMed, DOI, ClinicalTrials.gov).
Educational only, not medical advice. Grades and scores reflect published evidence weighted by study design and quality; see the methodology.