Metabolic Access

Tag: Clinical evidence

  • Does Eating Carbs Last Lower Blood Sugar? What Meal-Order Trials Show

    Published: October 25, 2026 · Last updated: October 25, 2026

    Quick answer

    Eating carbs last, after vegetables and protein, can reduce the early glucose rise in small human trials. However, a flatter first-hour curve does not establish sustained weight loss, diabetes prevention, or a large HbA1c improvement. Recent reviews reach different estimates, and the choice of comparison and measurement time matters. Meal order is a possible addition to a balanced eating plan, with treatment decisions still guided by your care team. 2017 controlled trial, 2026 vegetable-preload review.

    Imagine a plate of chicken or tofu, vegetables, and rice. The question is whether beginning with the vegetables and protein changes the response to that same meal. That differs from adding a protein shake, removing the rice, or eating less overall. These approaches change different variables, so their results should not be treated as interchangeable.

    The points worth keeping

    • The strongest signal is short term. Several small crossover trials find a lower early glucose rise when carbohydrate comes later.
    • The comparator changes the story. Beating bread or rice eaten first is not necessarily the same as beating an ordinary mixed meal.
    • A hormone change is not a weight-loss result. Glucose, appetite, food intake, and long-term weight need separate measurement.
    • Longer-term evidence is less decisive. One newer meta-analysis reports a small HbA1c benefit; an earlier review and a 16-week pilot offer more cautious findings. The reviews overlap and cannot be added together as independent evidence. Updated type 2 diabetes review, 16-week pilot.

    What exactly is “carbs last”?

    In these studies, “carbohydrate” usually means the concentrated starch or sugar portion: bread, rice, or juice. Vegetables and protein foods can contain carbohydrate too. The label is a practical shorthand, not a claim that the first part of the meal is carbohydrate-free.

    Some protocols separate courses with a timed break. Others change the sequence without a pause. Neither approach establishes that everyone needs a stopwatch at dinner. The useful research question is whether the same foods produce different outcomes when their order changes.

    Trials in type 2 diabetes and prediabetes

    On a small screen, swipe the tables sideways to read every column. The “±” values below are identified as standard error (SE) or standard deviation (SD); they are not confidence intervals.

    Meal-order trials in diabetes and prediabetes
    Study and designPeople and comparisonDuration and resultWhat this establishes—and does not
    Shukla 2017; randomized crossover16 adults with metformin-treated type 2 diabetes; same meal, carbohydrate last vs first or togetherThree test days, glucose sampled for 180 minutes. Incremental peak: 34.7 ± 4.1 vs 75.0 ± 6.5 mg/dL, last vs first (mean ± SE; P < .001)A lower rise above baseline in this meal experiment; not a 54% reduction in total blood glucose or diabetes risk. One participant lacked mixed-meal data. Selected original methods/results checked. PMID 28989726
    Shukla 2019; randomized crossover15 participants with prediabetes; vegetables plus protein first, vegetables first, or carbohydrate firstThree test days, 180-minute sampling. Incremental peaks were over 40% lower in both vegetable-first conditions than carbohydrate firstSupports an acute response in prediabetes; no long-term prevention or weight outcome. Original abstract checked; full text not verified. PMID 30101510
    Touhamy 2025; within-person crossover20 adults with metformin-treated type 2 diabetes; provided meals, carbohydrate last vs first, without a rest intervalSix days per sequence. Free-living time in range: 84.8% vs 78.6% (P = .041); average glucose: 146.0 vs 152.3 mg/dL (P = .202)Short-term variability improved; average glucose difference was not statistically significant. Provided meals limit generalization. Indexed original methods/results checked; complete article extraction unavailable. PMID 39688621

    Read the glucose metric before reading the percentage

    An incremental peak is the increase above the premeal value. A percentage reduction in that increase can look much larger than the percentage change in the absolute glucose reading. Likewise, incremental area under the curve, or iAUC, summarizes the rise above baseline over a specified interval. It is not interchangeable with the highest reading or a single two-hour measurement.

    The 2025 experiment is useful because it tested sequencing without a prescribed pause. Yet its 12-day crossover observation cannot answer whether the habit remains useful for a year. A significant result for one glucose metric also does not make every other glucose metric significant. Original 2025 report.

    A 2026 finding: the first hour and the full three hours can disagree

    Newer trials: glucose outcomes and their boundaries
    Study and designPeople and comparisonResult and uncertaintyInterpretation limit
    Choi 2026; four-condition randomized crossover33 analyzed healthy Korean women with normal fasting glucose; vegetables–protein–carbohydrate, reverse order, or meals eaten together; three consecutive test days per conditionTable 2: 0–3-hour glucose iAUC was 146.63 ± 10.99 vs 148.59 ± 10.99 mmol·min/L for vegetables first vs together (mean ± SE; no significant pairwise difference). Early 0–2-hour iAUC favored vegetables firstEarly benefit did not mean lower total three-hour exposure than eating together. Figure 1 reports 36 randomized; the participant-text denominator is inconsistent, so 33 analyzed is used here. Selected full-text methods/Table 2 checked. DOI 10.1111/dom.71227
    Giro 2026; randomized crossover14 highly trained male soccer players; standardized breakfast, carbohydrate last vs firstIncremental glucose peak difference: −1.7 mmol/L (reported interval −2.4 to −1.0; P < .001). Greater reported hunger, lower satiety; no difference in subsequent 3–6-hour unrestricted intakeLower glucose did not establish better appetite control or reduced eating. Small athletic population; original abstract checked. PMID 42617805

    Match the claim to the population and outcome

    The Korean trial also found that absolute glucose peaks were similar for vegetables-first and together conditions. Its title describes a high-risk phenotype, but eligibility excluded diabetes and prediabetes. It should not be presented as a treatment trial in people with either diagnosis. The study was published online in August 2026, ahead of its November issue. Original study, Methods and Table 2.

    These results make “Did it lower the spike?” an incomplete question. Ask which comparison, which outcome, and how long the investigators watched. In the athlete trial, a lower glucose peak did not translate into lower measured food intake later that day. That is a useful caution against treating one favorable metabolic number as a summary of the whole response.

    Does eating carbs last improve HbA1c or help with weight loss?

    The most directly practical pilot randomized 45 adults with overweight or obesity and prediabetes to nutrition counseling alone or counseling plus food-order advice. 39 completed 16 weeks: 18 in food order and 21 in control.

    Weight changed by −3.6 ± 5.7 vs −2.6 ± 6.8 lb (mean ± SD; between-group P = .625). HbA1c changed by −0.10 ± 0.20 vs −0.03 ± 0.30 percentage points (between-group P = .364). These were not convincing between-group advantages. A significant change inside one group, by itself, does not show that intervention beat control. Attrition, small sample size, and pandemic disruption limit precision. Original pilot, Results and Table 2.

    Why reviews reach different conclusions

    Published reviews: scope, estimates, and limits
    Review—not an additional trialScopeKey resultBoundary
    Okami 2022Eight trials/230 participants in quantitative synthesis overall; HbA1c used three trials/147 participantsHbA1c mean difference −0.21 percentage points; 95% CI −0.44 to +0.03; low certaintyInterval includes no effect; search ended April 2020. Selected original results/Table 2 checked. PMID 35210284
    Saldarriaga-Callejas, online 2025; 2026 issue17 studies/389 participants overall, adults with type 2 diabetesHbA1c difference −0.16 percentage points; 95% CI −0.31 to −0.01; P = .04Small pooled benefit, not a large effect. Outcome-specific denominator and all trial estimates not re-extracted. Original abstract checked. Publisher history identifies the 2026 correction as a missing author. Original review and history
    Kampmann-Gongoll 202618 randomized trials; vegetable preload, adults with and without health conditionsGlucose difference at 30 minutes −1.21 mmol/L (95% CI −1.75 to −0.68); at 120 minutes +0.24 mmol/L (+0.06 to +0.42); very low certaintyDifferent intervention scope and time points. Original abstract checked; no fresh pooled HbA1c estimate extracted. PMID 42748597

    What the disagreement means for the conclusion

    The newer type 2 diabetes review supports the possibility of a small HbA1c improvement. It should be included rather than allowing the older null estimate to stand as the final word. However, the newer vegetable-preload review shows why early and later glucose values may point in opposite directions. Its abstract also gives a before-and-after HbA1c example; that is not a stated pooled between-group treatment effect.

    Different eligibility rules, preloads, mixed-meal comparators, and endpoints can yield different estimates. We have not independently reconciled every included trial or re-run these analyses. Their study counts overlap; summing them would exaggerate the evidence base. Taken together, they support a cautious distinction: promising short-term glucose effects, with a smaller and less secure long-term benefit.

    How to try meal order without turning dinner into a test

    If this fits your medical plan, use foods already in the meal. The following examples are practical adaptations, not menus tested head to head in these trials:

    1. Build a balanced plate first. Include nonstarchy vegetables, a suitable protein food, and a carbohydrate portion that fits your needs. CDC’s plate method is a starting framework. CDC meal planning.
    2. Begin with some vegetables and protein. For a rice plate, that might mean broccoli and tofu before starting the rice. For a sandwich lunch, an available vegetable side is an option; dismantling every sandwich is unnecessary.
    3. Keep portions and drinks in view. A change in order does not make a large sugary drink nutritionally equivalent to water. Adding extra cheese, oil, or a shake also changes the meal, so it no longer tests order alone.
    4. Choose a habit you can maintain comfortably. Food order is optional. An ordinary mixed meal is not a failure, and a meal with beans or lentils need not be sorted into artificially separate nutrients.
    5. Evaluate the outcome that matters. If you have diabetes, discuss your clinician-set glucose goals and follow-up results. If the goal is weight management, look at sustained intake, nutrition, and weight—not just the most flattering CGM screenshot.

    For related evidence, see our reviews of walking after meals, CGM for weight loss without diabetes, and fiber and metabolic outcomes. These are separate interventions; their effects cannot simply be added together.

    When a meal-order experiment needs medical guidance

    People using insulin or medicines that can cause hypoglycemia should discuss meaningful changes in carbohydrate timing or quantity with their care team. Skipping or delaying food can contribute to low glucose with these treatments. Do not independently change a dose to match this article. NIDDK hypoglycemia guidance.

    Treat low glucose promptly according to your care plan. Do not eat vegetables first to “flatten” the response to rescue carbohydrate. Fast-acting carbohydrate is used because glucose needs to rise quickly; fiber and fat can delay absorption. CDC low-glucose treatment.

    These small adult studies do not establish a universal routine for children, pregnancy, or people with complex digestive or nutritional needs. Discuss the fit with a clinician or dietitian rather than transferring a study protocol directly to your situation.

    Frequently asked questions

    Must I wait exactly 10 minutes before eating rice or bread?

    No universal waiting time is established. Some early studies imposed a rest interval; the 2025 study tested no rest interval. This supports flexibility, not a claim that every sequence or waiting time is equally effective. 2025 trial.

    Is this a natural substitute for a GLP-1 medicine?

    No. A change in a meal-related gut-hormone response does not establish the sustained exposure, clinical outcomes, or treatment equivalence of a medicine. The food-order studies reviewed here do not test replacement of prescribed GLP-1 therapy.

    Do I need a CGM to do this?

    No device purchase is required to begin a balanced meal with vegetables. Whether CGM is useful is a different clinical question. A lower reading in a personal comparison does not establish long-term benefit; our CGM evidence review addresses that distinction.

    Does this mean people without diabetes should avoid normal glucose rises?

    The trials show differences between meal conditions. They do not establish that every rise after eating is harmful or that healthy people should pursue the flattest possible curve. Avoid turning an experimental marker into a personal diagnosis.

    How this review was prepared—and its limits

    On October 25, 2026, we compared the public inventories of Aether Health (54 posts) and Metabolic Access (30). No existing English article focused on meal order. The Korean article supplied the topic; this English version was independently expanded with newer trials and conflicting reviews, rather than reproducing its claims or reviewer credit.

    Targeted web searches used “food order,” “carbohydrate last,” “meal sequence,” “randomized,” and “HbA1c,” including 2025–2026 updates and linked correction checks. Original records were checked through PubMed/Europe PMC, PMC and publisher pages. Selection emphasized human comparisons, population differences, short versus longer outcomes, and conflicting results. This is a selective narrative review, not an exhaustive systematic search, meta-analysis, or independent duplicate extraction.

    Selected original full-text passages/tables were checked for 2017, the 2023 pilot, the 2026 Korean trial, and the 2022 review. The 2019 and athlete studies and the newer meta-analyses were assessed at original-abstract level. The 2025 study’s indexed original methods/results were available, while direct full-text access failed. Missing fields remain missing. The Korean report’s recruitment wording conflicts with its flowchart; analyzed N = 33 is clear, and its title should not override its normal-glucose eligibility criteria.

    The Korean 2026 study reports Pulmuone Foundation program support and no declared conflicts. The 2023 pilot reports author equity/consulting relationships. Funding and conflicts were not independently audited across every study. No article-specific scientific advisory review has been confirmed. AI tools assisted drafting and source organization.

    Sources and further reading

    Written and edited by: Denny Jaeyong Choi (JD), INSEAD MBA

    Update history: October 25, 2026—first English publication; added 2025–2026 evidence, early-versus-late glucose comparisons, the 16-week null between-group findings, review disagreements, and explicit access limits. Publication preparation did not change the clinical conclusions or add a scientific-review attribution.

    Medical information: This article provides general education and does not diagnose, prescribe, or replace individualized medical advice.

  • Intermittent Fasting vs Calorie Restriction: What Human Trials Actually Test

    Quick answer

    Intermittent fasting can be a useful way to organize eating, but it is not one treatment. A daily eating window, three low-calorie days per week, and an early dinner are different interventions. In selected randomized trials, adding an eating window did not clearly improve weight loss when calories were controlled. A separate 4:3 fasting program did outperform daily restriction, but both groups received intensive support and actual food intake was not experimentally held equal. Small metabolic studies also suggest that weight and glucose-related outcomes need separate answers. Isocaloric trial, 12-month calorie-restriction trial, 4:3 trial.

    Published: October 24, 2026 · Last updated: October 24, 2026 · Initial source search: October 22, 2026

    The key takeaways

    • Ask what fasting was compared with: usual eating, an equally supported diet, or food provided at matched calories.
    • A useful eating routine need not have a unique calorie-independent weight-loss effect.
    • Better insulin sensitivity, a small HbA1c change, and sustained weight loss are not interchangeable outcomes.
    • A fasting schedule used inside a supervised trial is not a universal self-treatment prescription.

    The practical question is not whether the clock matters at all. It is whether changing the clock adds something beyond changing the food, calories, and support around it. That distinction makes apparently conflicting headlines easier to read.

    First, separate three different questions

    Can this routine help someone eat differently? A shorter eating window might change opportunities to snack or make meal planning simpler. A real-world trial can test the whole strategy without proving which component caused the result.

    Does timing change weight when food intake is held equal? That requires a more controlled comparison. Prescribing the same calorie target is not the same as providing and matching the calories people actually consume.

    Can timing affect a metabolic marker without weight loss? This is a third question. An answer about insulin sensitivity cannot automatically become an answer about pounds lost, diabetes prevention, or lifespan. Our review of metabolic health and aging explains the same distinction between biomarkers and clinical outcomes.

    Keep those questions beside you when reading the studies below. They are a comparison of selected trials, not five votes on one identical intervention.

    Five trials, five different comparisons

    On smaller screens, swipe or scroll sideways within each table to read every column.

    Timing with matched food or prescribed calorie targets

    Selected human trials: populations, comparisons, outcomes, and boundaries
    Trial and sourcePopulation, denominator, and durationIntervention and comparatorResult and uncertaintyWhat it can—and cannot—answer
    Maruthur, 2024. PMID 38639542; DOI 10.7326/M23-3132. Original abstract41 randomized adults with obesity and prediabetes or diet-controlled diabetes; all completed 12 weeks.Provided isocaloric diets: 10-hour window, 80% of calories before 1 p.m., versus a usual pattern of up to 16 hours, at least 50% after 5 p.m.Weight change −2.3 versus −2.6 kg; between-group difference +0.3 kg, 95% CI −1.2 to +1.9. Glycemic changes did not differ.No demonstrated extra weight benefit in this small controlled setting; not proof that every schedule is equivalent.
    Liu, 2022. PMID 35443107; DOI 10.1056/NEJMoa2114833. Original abstract139 randomized adults with obesity; 118 completed the 12-month visit.Eating 8 a.m.–4 p.m. plus prescribed calorie restriction, versus prescribed calorie restriction alone.Weight change −8.0 versus −6.3 kg; adjusted net difference −1.8 kg, 95% CI −4.0 to +0.4; P=.11.No statistically clear added weight benefit. Prescribed targets do not establish identical consumed calories.

    Weekly fasting within a supported weight-loss program

    Selected human trials: populations, comparisons, outcomes, and boundaries
    Trial and sourcePopulation, denominator, and durationIntervention and comparatorResult and uncertaintyWhat it can—and cannot—answer
    Catenacci, 2025. PMID 40163873; DOI 10.7326/ANNALS-24-01631. Original abstract165 randomized adults, age 18–60, BMI 27–46; 84 versus 81 allocated; 125 completed 12 months.80% energy restriction on three nonconsecutive days versus 34% daily restriction; both received a high-intensity behavioral program.Intention-to-treat result: 2.89 kg greater loss with 4:3 fasting; 95% CI for the greater-loss magnitude 0.14–5.65 kg; P=.040.A positive program-level comparison, not proof of extra loss at identical actual calorie intake or of a daily 16:8 schedule.

    Metabolic markers, not extra weight loss

    Selected human trials: populations, comparisons, outcomes, and boundaries
    Trial and sourcePopulation, denominator, and durationIntervention and comparatorResult and uncertaintyWhat it can—and cannot—answer
    Sutton, 2018. PMID 29754952; DOI 10.1016/j.cmet.2018.04.010. Original abstract and figure captionsEight male completers with prediabetes; crossover, five weeks per schedule.Weight-maintaining controlled feeding: six-hour early window, dinner before 3 p.m., versus 12-hour window.Improved insulin-sensitivity-related measures; glucose during the oral glucose tolerance test was not improved. Effect-size CIs were not extracted for this review.A small proof-of-concept metabolic signal, not evidence of sustained weight loss or diabetes prevention.
    TIMET, 2024. PMID 39348690; DOI 10.7326/M24-0859. Original abstractAdults with metabolic syndrome and elevated glucose or HbA1c; medication allowed. 108 completers, reported as 89% of randomized participants; three months.Standard nutritional counseling versus the same care plus a personalized 8–10-hour window and app-based timing records.Between-group HbA1c difference −0.10 percentage points, 95% CI −0.19 to −0.003; age-adjusted.A modest glycemic signal on top of usual care; self-reported diet and multiple changing elements prevent clean calorie-independent attribution.

    Reading the numbers: A negative weight difference favors greater loss only when the comparison is defined that way. The 4:3 row instead states a positive magnitude of greater loss. A confidence interval crossing zero does not establish an added benefit; it also does not prove exact equality. The reported adjusted Liu estimate need not equal subtraction of rounded group means.

    Why the controlled-feeding result matters

    The 2024 feeding trial addresses timing more directly than advice alone: calories and nutrient content were matched. It did not demonstrate an added weight benefit. Generalizability remains limited: 93% of participants were women and 93% were Black. Maruthur trial.

    Do not reverse that finding into “the clock never matters.” A controlled experiment asks a narrower question than whether a routine is convenient, culturally workable, or easier for someone to follow. Mechanistic specificity and practical usefulness are separate tests.

    Why the positive 4:3 trial is not a contradiction

    The 2025 study compared supported weekly restriction patterns, not identical provided meals. Its positive result supports that program; it does not isolate a calorie-independent mechanism. Catenacci trial.

    A positive program result should not be dismissed merely because behavior may contribute. Behavior is part of a lifestyle intervention. Equally, a successful program should not be marketed as proof of a unique fasting mechanism. Both statements can be true.

    Can glucose regulation improve without more weight loss?

    The small early-feeding crossover study provides a reason not to answer every question with the scale. Its insulin-related signal did not mean all glucose measures improved. TIMET, meanwhile, reported a modest HbA1c improvement in a different population receiving usual clinical care. Neither establishes a universal eating window or a reduction in future clinical events. Sutton, TIMET.

    Also, −0.10 percentage points is not a 10% reduction. Units matter, particularly when a small average difference is repackaged as a dramatic headline. For another example of keeping glucose feedback separate from weight outcomes, see our CGM evidence review.

    What newer follow-up evidence adds

    A 2026 Clinical Nutrition report describes a 99-person follow-up cohort after a 12-week early, late, or self-selected eating-window intervention, with Mediterranean-diet education in all groups. Its indexed publisher abstract reports maintained weight differences for early and late schedules at a 12-month post-intervention assessment, but only 26% reported using time-restricted eating during follow-up. That raises useful questions about sustained behavior rather than settling a timing-only mechanism. Publisher abstract, DOI 10.1016/j.clnu.2026.106706.

    We did not verify the complete follow-up results, effect-size confidence intervals, or reconciliation with the parent cohort. We therefore discuss it separately from the main five-trial comparison, do not rank early against late schedules, and do not treat the follow-up as an independent new randomized population. Long-term maintenance deserves its own question, as our Akkermansia maintenance review also emphasizes.

    A practical checklist before choosing a routine

    Use this as preparation for a discussion, not a prescription to copy an extreme trial schedule.

    1. Define the goal. Write down whether you want simpler meal planning, weight change, or a clinical glucose target. Decide which outcome would actually answer your question.
    2. Describe the existing pattern. Note the first and last caloric intake, work hours, and medicines that require food. Do not assume everyone begins with the long eating window used in some studies.
    3. Check safety first. Discuss diabetes medicines, pregnancy or breastfeeding, eating-disorder history, and other conditions before restricting meals. NIH and NIDDK emphasize individualized advice; do not change doses yourself. NIH, NIDDK, NHS-linked patient guidance.
    4. Agree on how to evaluate it. Keep nutrition, feasibility, and the chosen outcome in view. Recording everything is not necessary for everyone; avoid turning a routine into a rigid test of willpower.
    5. Review problems instead of escalating restriction. Ask for help if the schedule creates distress, interferes with treatment, or leaves you unable to eat adequately. A shorter window is not automatically a better intervention.

    Frequently asked questions

    Is 16:8 fasting better than counting calories?

    The answer depends on the actual comparison. The selected trials do not establish that every daily fasting window beats an equally supported calorie-restriction strategy. Nor can the positive 4:3 program be substituted for a 16:8 result. Choose evidence that matches the schedule and outcome being discussed. Liu, Catenacci.

    Does eating earlier burn more fat at the same calories?

    That is not established by the selected isocaloric result. More generally, temporary fuel use and sustained weight change are different questions; evidence for one should not silently answer the other. Maruthur.

    Can I stop diabetes medication if my readings improve?

    No change should be made on the basis of this article. Fasting can complicate medication management, including hypoglycemia risk with insulin and some other treatments. Plan any meal-pattern change with the prescribing clinician rather than skipping or adjusting medication yourself. NIDDK.

    Is this an anti-aging intervention?

    Weight, HbA1c, and insulin-related measures are not lifespan. These selected trials do not demonstrate longer human life from a fasting schedule. Biological plausibility should generate better research questions, not become a promise of age reversal. See our separate human aging-trial review.

    Is a CGM needed to decide whether fasting works?

    A favorable sensor trace cannot by itself establish sustained weight benefit, and wearing a device is not equivalent to receiving structured guidance. The relevant outcome and clinical need come first. Our CGM review examines those separate questions; this article does not recommend buying a device.

    Methods, access limits, and update policy

    On October 22, 2026, we checked the site’s 29 published titles to avoid a duplicate review. We used targeted web searches for randomized isocaloric feeding, calorie restriction with time-restricted eating, the TIMET trial, and the 4:3 program, then checked original PubMed abstracts and the Sutton figure captions. Liu’s original indexed abstract was also retrieved through Europe PMC. We screened the 2026 publisher-indexed follow-up summary separately.

    This is a narrative comparison, not an exhaustive search, systematic review, or meta-analysis. It deliberately contrasts designs rather than pooling different interventions. We did not independently extract complete protocols, supplements, endpoint hierarchies, every subgroup, or all disclosure forms. The TIMET randomized denominator is not reconstructed from a rounded completion percentage. Full-text attempts that returned errors are not counted as successful access. Newer or unlocated studies may change the interpretation.

    Funding and individual investigator relationships can influence interpretation and warrant further checking; the original records link disclosure information, but we did not audit all forms. AI tools assisted drafting and source organization; no new independent scientific or medical review is claimed. Future substantive changes will identify the source, changed interpretation, and reason—not merely refresh the date.

    Sources and further reading

    • Maruthur NM et al. Isocaloric time-restricted eating. Annals of Internal Medicine, 2024. PMID 38639542; DOI 10.7326/M23-3132.
    • Liu D et al. Calorie restriction with or without time-restricted eating. NEJM, 2022. PMID 35443107; DOI 10.1056/NEJMoa2114833.
    • Catenacci VA et al. The 4:3 intermittent-fasting program. Annals of Internal Medicine, 2025. PMID 40163873; DOI 10.7326/ANNALS-24-01631.
    • Sutton EF et al. Early time-restricted feeding. Cell Metabolism, 2018. PMID 29754952; DOI 10.1016/j.cmet.2018.04.010.
    • Manoogian ENC et al. TIMET. Annals of Internal Medicine, 2024. PMID 39348690; DOI 10.7326/M24-0859.
    • 2026 post-intervention follow-up, limited to indexed abstract access. Clinical Nutrition; DOI 10.1016/j.clnu.2026.106706.
    • NIH: To Fast or Not to Fast; NIDDK: Fasting Safely with Diabetes.

    Written and edited by: Denny Jaeyong Choi (JD), INSEAD MBA

    Editorial history: October 22, 2026 — initial draft prepared; comparisons and access limitations recorded. October 24, 2026 — prepared for publication after restoring editor access; publication labels and layout finalized. This is not a new systematic search or independent scientific review.

    Medical information notice: This article provides general education, not personal diagnosis or treatment. It does not prescribe a fasting schedule or medication adjustment. Consult a qualified health professional about your circumstances.