The phases of a fast: what happens in the body, and why

Phase 1 · roughly the first 2 to 5 hours after a meal

Fed, or absorptive

What happens

• During the absorptive phase, insulin is high and the liver takes glucose up rather than putting it out, storing it as glycogen and converting part of the surplus into fat. [1]
• Insulin also restrains the release of fat from adipose tissue, so while a meal is being absorbed the body is running mostly on what has just been eaten rather than on its own stores. [1]·[3]
• The stomach empties gradually rather than all at once. In 73 healthy volunteers given a standardised test meal, the median half-emptying time was 44 minutes for the liquid component and 162 minutes for the solid component, with reference intervals of 28 to 78 and 144 to 193 minutes respectively. [2]

Why it happens

• Insulin is the signal that fuel has arrived, and its actions are coordinated rather than separate: the same hormone switches on storage and switches off release, while glucagon does the reverse. That single opposition is what drives every phase that follows. [1]
• A fast does not begin when you put your fork down; it begins when absorption finishes. Because gastric emptying varies severalfold between healthy people and depends on what the meal contained, the end of the fed state is a range rather than a time. [2]

What it feels like

Commonly reported experience, not measured outcomes. Where a symptom has actually been studied it appears above or below with its source.

• —Fullness, then a slow return to feeling neutral.
• —Hunger reappearing three or four hours after eating is ordinary and is not a sign that anything metabolic has happened yet.

What the evidence does not support

• The figure of "0 to 4 hours" for the fed state is a convention rather than a measurement, and gastric emptying data show why: the reference interval for the solid component of a standardised meal alone spans 144 to 193 minutes across healthy adults, before any real meal’s size, fat or fibre content is taken into account. [2]
• No study establishes an hour at which the fed state ends and a fast begins. Any page that gives one is reporting a convention, and the honest version of the claim is that absorption tails off rather than stopping. [2]

Phase 2 · roughly 4 to 16 hours

Post-absorptive

What happens

• Once absorption finishes, insulin falls and glucagon rises, and the liver reverses direction: it stops taking glucose up and starts releasing it. [1]
• Two processes supply that glucose at once. Glycogenolysis breaks down the liver’s stored glycogen, and gluconeogenesis builds new glucose from lactate, glycerol and amino acids. [4]·[1]
• Gluconeogenesis is already the larger contributor within the first day, not a later substitute for glycogen. Measured by 13C NMR through a 68-hour fast, gluconeogenesis accounted for 64 ± 5% of all glucose production across the first 22 hours; measured independently with deuterated water, it accounted for about half at 14 hours and about two-thirds at 22 hours. [4]·[5]
• Falling insulin releases the brake on adipose tissue, so free fatty acids appear in the blood in growing quantity and take over an increasing share of the body’s fuel supply. [1]·[3]

Why it happens

• The liver’s glycogen store is finite and the brain’s demand for glucose is not, so gluconeogenesis has to run alongside glycogenolysis from early on rather than waiting for the store to empty. That is the direct explanation for why the isotope measurements find it contributing the majority of glucose within the first day. [4]·[5]
• Insulin is the brake on fat release, and when insulin falls the brake comes off. The fatty acids that appear in this phase are the raw material for everything later in a fast, including every ketone the liver eventually makes. [1]·[3]

What it feels like

Commonly reported experience, not measured outcomes. Where a symptom has actually been studied it appears above or below with its source.

• —For most people, nothing in particular. This is the ordinary overnight state, and reaching it is not an achievement.
• —Hunger tends to arrive in waves around the times you normally eat rather than climbing steadily.

What the evidence does not support

• The widely repeated sequence — the body burns through glycogen first and only then switches to making glucose — is not what the human measurements show. Both isotope methods find gluconeogenesis supplying roughly half or more of glucose production within the first day of a fast. [4]·[5]
• The hunger claim has a weaker source than it is usually given. The observation that ghrelin keeps a meal-related rhythm during a fast, with roughly eight pulses in 24 hours and an overall downward drift, comes from six volunteers in a study its own authors call a pilot — and it measures a hormone, not hunger. [13]
• Both glucose-production studies are small, in healthy adults, under controlled laboratory conditions. They settle where circulating glucose comes from. They are not health outcomes and they do not describe anything a person can feel. [4]·[5]

Phase 3 · roughly 16 to 24 hours

Early fasting

What happens

• Gluconeogenesis becomes overwhelmingly the source of blood glucose as this window passes. Measured by 13C NMR, it supplied 82 ± 5% of glucose production over the period from 22 to 36 hours of fasting and 96 ± 1% from 36 to 54 hours; measured independently with deuterated water, it supplied 93 ± 2% at 42 hours. [4]·[5]
• The liver converts a growing share of the fatty acids reaching it into ketone bodies, whose production rate and blood concentration rise steeply through the early phase of a fast and only begin to plateau after about five days. [1]·[6]
• Resting energy expenditure does not fall in this window. In 11 healthy lean adults fasted for 84 hours, resting energy expenditure measured by indirect calorimetry was significantly higher on day 3 than on day 1, while noradrenaline roughly doubled by day 4. [10]
• Fasting causes real loss of sodium in the urine, which begins early and peaks within the first days of a fast. [27]·[28]

Why it happens

• Low insulin and high glucagon switch the liver from storing fatty acids to burning them, and ketogenesis is what the liver does with more fatty acid than it can oxidise for its own needs. Ketones are the export form of fat-derived energy. [1]
• The rise in metabolic rate is not mysterious. Noradrenaline roughly doubled over the same period, and the authors of that study propose the falling blood glucose as the trigger and the catecholamine rise as the initial signal for the metabolic changes of early starvation. [10]
• Sodium is lost because the hormonal shift of fasting changes how the kidney handles it, which is why light-headedness on standing during a fast is often a salt and water problem rather than a blood sugar one. [27]·[28]

What it feels like

Commonly reported experience, not measured outcomes. Where a symptom has actually been studied it appears above or below with its source.

• —This is where most people first notice something.
• —Hunger, irritability, feeling cold, and light-headedness on standing are commonly reported.
• —Headache is the best-documented complaint of the three, and caffeine withdrawal is a plausible cause of it if you normally drink coffee.
• —Concentration is reported both ways — some people find it better, some worse — and there is no good human trial settling which is typical.

What the evidence does not support

• Sixteen hours is the most repeated number in fasting content and it is a schedule, not a physiological threshold. The isotope measurements describe a continuum with nothing switching at 16 hours, and the popular figure comes from the 16:8 eating pattern rather than from any measurement. [4]·[5]
• The claim that metabolism "slows down" or enters starvation mode within a day or two runs in the opposite direction to the only direct measurement of it: resting energy expenditure rose. That study is 11 lean people over 84 hours, and it says nothing about the metabolic adaptation that does occur over weeks or months of sustained energy restriction. [10]
• Fasting headache has been reviewed but not explained. The review most often cited concludes that fasting headache usually resembles tension-type headache and becomes more likely the longer the fast continues, with hypoglycaemia and caffeine withdrawal implicated as causes — implicated, not demonstrated. [14]
• This is the window most often claimed for autophagy, and no human study establishes a threshold hour for it. That claim is dealt with separately below rather than repeated per phase. [17]·[16]

Phase 4 · roughly 24 to 48 hours

Ketosis and the second day

What happens

• Ketone concentrations continue to climb through this window and do not plateau until about the fifth day of a fast. Part of the rise is falling clearance rather than rising production: as blood ketones increase, muscle uptake saturates and ketones are taken up preferentially by the brain instead. [6]
• Growth hormone secretion rises sharply. In nine healthy men sampled every five minutes for 24 hours, the second day of a fast produced a fivefold increase in total growth hormone production, through both more secretory bursts — 32 a day against 14 — and larger ones. [11]
• Protein breakdown increases rather than decreases across the first days of a fast. After 60 hours without food, urinary nitrogen loss rose, whole-body leucine flux and oxidation rose by 50 to 75%, and net amino acid release from forearm muscle was two to three times its 12-hour value. [12]
• The brain does start using ketones in this period, and the measured contribution is partial. After 3.5 days of starvation in nine healthy volunteers, cortical glucose metabolism fell by 26% and net beta-hydroxybutyrate uptake rose thirteenfold, with ketones supplying approximately one quarter of the brain’s energy requirement. [7]

Why it happens

• Ketones exist because the brain cannot run on fatty acids in useful quantity, and the liver therefore converts fat into a fuel that crosses into the brain and can replace part of its glucose demand. That substitution is what makes prolonged human fasting survivable at all. [9]·[1]
• Growth hormone rises because its pulses come more often and carry more hormone, a pattern consistent with more frequent releasing-hormone signals and longer withdrawal of the inhibitory signal that normally restrains them. [11]
• Protein is broken down because gluconeogenesis needs carbon, and once glycogen is spent and glycerol is limited the remaining source is amino acid taken from the body’s own tissue. The nitrogen appearing in urine is the accounting record of that. [12]·[9]

What it feels like

Commonly reported experience, not measured outcomes. Where a symptom has actually been studied it appears above or below with its source.

• —Hunger is frequently reported as easing rather than intensifying after the first day, which surprises most people the first time.
• —Feeling cold, low tolerance for exercise, disturbed sleep and a distinctive breath are all commonly reported at this point.
• —Some people feel unusually clear-headed and some feel flat. Both are reported often enough that neither should be presented as the normal response.

What the evidence does not support

• Does the brain run on ketones after 48 hours of fasting?

  The claim that the brain "runs on ketones" by 48 hours is not supported by any measurement at 48 hours. The study behind it examined three patients after five to six weeks of starvation, in whom ketones had indeed replaced glucose as the predominant cerebral fuel. The direct short-fast measurement, at 3.5 days in nine volunteers, put ketones at about one quarter of the brain’s energy — and 3.5 days is already past the window the claim is made about. [8]·[7]
• Protein sparing is being borrowed from the wrong timescale. Over weeks of starvation nitrogen loss does fall from its early peak, and that is the classic observation. In the first days it moves the other way: nitrogen loss and muscle amino acid release both increased at 60 hours. A page saying the body begins sparing protein at 24 hours is describing a multi-week curve and printing it on day two. [9]·[12]
• The growth hormone rise is a hormone measurement and not an outcome. IGF-I, the mediator through which growth hormone does most of its anabolic work, was unchanged after 56 hours of fasting in the same study, and no trial shows that the growth hormone rise of a short fast produces muscle, fat loss, or any other result. [11]
• Almost every study describing this window is very small, because the measurements require catheters, biopsies or five-minute blood sampling: three patients, nine volunteers, nine men, eight subjects, eleven adults, five adults. That is the honest state of the evidence, and it is a reason to treat every number above as a direction rather than a value. [8]·[7]·[11]·[12]·[10]·[15]
• A fast perturbs routine blood tests. In five healthy adults fasted for 72 hours, total cholesterol, C-reactive protein, TSH and free T3 all shifted by 72 hours and recovered after refeeding — which matters mostly as a warning that a blood test taken during an extended fast is not measuring your baseline. [15]

Autophagy, and the hour nobody has measured

• The clearest direct comparison ran mice and humans side by side: intermittent fasting raised autophagy markers in mouse liver, did not raise them in mouse muscle, and in human skeletal muscle a 24-hour fast moved several markers in the opposite direction to the popular claim. [16]
• A 36-hour human fast with serial muscle biopsies at 2, 12, 24 and 36 hours found autophagy, in the authors’ own summary, "only modestly affected", with the response differing according to whether participants were trained or untrained. [17]
• A 72-hour human fasting study states plainly that autophagy was not measured directly and that only an environment conducive to it was inferred — which is the position most fasting content quietly occupies without saying so. [15]

The benefits, with their mechanisms and their limits

A benefits list without mechanism is what every content farm already publishes, and a benefits list without exclusions is marketing. Each entry below gives the proposed mechanism, the strength of the human evidence, and who it does not apply to.

Salt and water, the one mistake that matters

This is separated from the list below because it is not about who should avoid fasting. It is the one active mistake that turns an unpleasant fast into a dangerous one, and it is usually made by people who are doing everything else right.

• Fasting causes real loss of sodium in the urine, which starts early and peaks within the first days — so the light-headedness people attribute to low blood sugar during a fast is often a salt and water problem instead. [27]·[28]
• Drinking large volumes of plain water to blunt hunger during a fast is the one common practice with a serious downside, because sodium is already being lost and a large plain-water intake dilutes what remains. The same mechanism causes exercise-associated hyponatraemia, whose early signs — headache, nausea, confusion — are easy to mistake for ordinary fasting discomfort. [29]·[30]

Fasting Index does not tell you how much salt to take, because none of the studies above establishes a dose and inventing one would be worse than saying so. It sells no electrolyte product and recommends no brand.

Who should not fast

Not a footnote, and not conditional on how motivated you are. Each entry gives the reason and the source. If you are in more than one of these groups, the case for talking to a doctor first is stronger, not weaker.

• Anyone taking insulin or a sulfonylurea, and anyone with type 1 diabetes

  The risk is hypoglycaemia during the fast and ketoacidosis in type 1 diabetes, and both are managed by changing doses and monitoring glucose rather than by willpower. The International Diabetes Federation and the Diabetes and Ramadan International Alliance publish risk-stratified guidance covering exactly this, including for type 1 diabetes; the point of it is that the plan is made with the prescriber before the fast, not during it. [23]

• Anyone pregnant or breastfeeding

  The largest synthesis available covers Ramadan fasting — 22 studies and 31,374 pregnancies — and found no effect on birth weight and none on preterm delivery. It also found no data at all on perinatal mortality and concluded that the evidence on other outcomes is insufficient. An absence of detected effect on one outcome is not a finding of safety across all of them, and extended fasting is a different exposure from fasting between dawn and sunset. [24]

• Anyone with a history of, or currently experiencing, disordered eating

  In 491 adults followed for eight months, current or past engagement with intermittent fasting predicted higher eating-disorder psychopathology and greater eating-related psychosocial impairment at follow-up. The study is observational and cannot establish which came first, but an intervention whose entire structure is formalised restriction is a poor choice for someone with that history. [25]

• Anyone underweight, and children and adolescents

  The protein cost of the first days of a fast is measurable — nitrogen loss and muscle amino acid release both rose at 60 hours in healthy adults — and there is no reason to accept that cost in someone who has nothing spare or who is still growing. [12]

• Anyone whose medication has to be taken with food, or whose dose depends on eating

  This is a question for the prescriber and not for a website. It covers far more than diabetes medication, and the failure mode is silent: a dose that was correct with a meal is not automatically correct without one. [23]

• Anyone breaking a long fast after a period of undereating

  Refeeding causes electrolytes — phosphate in particular — to shift rapidly out of the blood and into cells as insulin rises again. The literature describing this is about critically ill hospital patients, so it is cited here for the mechanism rather than to imply that an ordinary fast in a well-nourished adult carries comparable risk. [26]

What this page does not cover

Stated rather than left as an absence. A reference that does not mark its own edges invites the reader to assume it has none.

• —Fasting beyond 48 hours, where the physiology changes again and supervision starts to matter.
• —Which fasting schedule to choose, or whether to fast at all.
• —Anything specific to Ramadan, which raises questions of ruling and of timing that this page is not competent to answer.
• —What to eat when you break a fast, beyond the warning about refeeding above.
• —Any claim that fasting treats, cures or reverses a disease. This site makes none.

Sources

30 primary sources. Every one was resolved against PubMed and its abstract read before it was entered here — none of it is cited from memory. Each carries the study design and sample it actually used, and a plain statement of what it cannot be used to show.

The samples are small throughout, and that is a property of the field rather than of this page: measuring hepatic glycogen, cerebral fuel use or muscle protein turnover in a fasting human requires spectroscopy, catheters or biopsies, so these studies are counted in single figures and tens of people.

1. [1]Energy metabolism in the liver — Compr Physiol, 2014PMID 24692138

   Design: Narrative review of hepatic fuel metabolism

   What it does not show: A review of mechanism drawing on human and animal work together. It describes how the liver switches between storing and releasing fuel; it measures nothing in any particular person and gives no timings.

2. [2]Clinical assessment of gastric emptying and sensory function utilizing gamma scintigraphy: Establishment of reference intervals for the liquid and solid components of the Nottingham test meal in healthy subjects — Neurogastroenterol Motil, 2017PMID 28589661

   Design: 73 healthy volunteers, stratified for age and sex, gamma scintigraphy

   What it does not show: The solid component of this test meal is twelve agar beads carrying no calories, and the liquid is a standardised 0.75 kcal/mL nutrient drink. It establishes how variable gastric emptying is between healthy people; it is not a measurement of how long your dinner takes to leave your stomach.

3. [3]Lipolysis during fasting. Decreased suppression by insulin and increased stimulation by epinephrine — J Clin Invest, 1987PMID 3540009

   Design: 13 volunteers, isotopic palmitate flux at 14 h and 84 h under a pancreatic clamp

   What it does not show: Measures fat release into the blood, not fat lost from the body. The clamp holds hormones constant precisely so the tissue response can be isolated, which is the opposite of ordinary fasting conditions.

4. [4]Quantitation of hepatic glycogenolysis and gluconeogenesis in fasting humans with 13C NMR — Science, 1991PMID 1948033

   Design: Serial 13C NMR of liver glycogen through a 68-hour fast, with tritiated glucose for whole-body glucose production

   What it does not show: Healthy adults under study conditions, and a small number of them. It settles where circulating glucose comes from and when; it says nothing about how anyone feels, and it is not a health outcome of any kind.

5. [5]Contributions of gluconeogenesis to glucose production in the fasted state — J Clin Invest, 1996PMID 8755648

   Design: Healthy subjects, deuterated water method, sampled at 14, 22 and 42 h of fasting

   What it does not show: An independent method reaching the same conclusion as the 13C NMR work, which is why both are cited. Small, healthy, controlled. The published variance around the 14 h and 22 h figures is wide, so treat those two as approximate and the 42 h figure as firm.

6. [6]Ketone body production and disposal: effects of fasting, diabetes, and exercise — Diabetes Metab Rev, 1989PMID 2656155

   Design: Review of human ketone turnover studies

   What it does not show: Describes the shape of the ketone curve and why it plateaus. It is a review, so it inherits the limits of the small turnover studies underneath it, and it reports no health outcome.

7. [7]Brain metabolism during short-term starvation in humans — J Cereb Blood Flow Metab, 1994PMID 8263048

   Design: 9 healthy volunteers, PET before and after 3.5 days of starvation

   What it does not show: Nine people, once, at three and a half days. It is the closest direct measurement available to the 24–48 h window and it is still past it, so it should be read as an upper bound on what ketones contribute at 48 h rather than a measurement of it.

8. [8]Brain metabolism during fasting — J Clin Invest, 1967PMID 6061736

   Design: 3 patients with obesity, cerebral vessel catheterisation after 5–6 weeks of starvation

   What it does not show: Three people, after five to six weeks without food. This is the study behind almost every "the brain switches to ketones" claim online, and it describes prolonged starvation, not a fast of one or two days.

9. [9]Fuel metabolism in starvation — Annu Rev Nutr, 2006PMID 16848698

   Design: Review, partly autobiographical, of four decades of human starvation research

   What it does not show: A synthesis by one of the field’s founders, describing prolonged starvation over weeks. Using its protein-sparing observation to describe the first 48 hours is the most common way this paper is misread.

10. [10]Resting energy expenditure in short-term starvation is increased as a result of an increase in serum norepinephrine — Am J Clin Nutr, 2000PMID 10837292

    Design: 11 healthy lean subjects, indirect calorimetry across 84 h of starvation

    What it does not show: Eleven lean people over three and a half days. It refutes the idea that metabolic rate falls within the first days of a fast. It says nothing about weeks of dieting, where the adaptation everyone means by "starvation mode" actually lives.

11. [11]Augmented growth hormone (GH) secretory burst frequency and amplitude mediate enhanced GH secretion during a two-day fast in normal men — J Clin Endocrinol Metab, 1992PMID 1548337

    Design: 9 normal men, blood sampled every 5 minutes for 24 h, fed day vs second day of a fast

    What it does not show: Nine men, and it measures a hormone rather than an effect. IGF-I — the mediator through which growth hormone does most of its anabolic work — was unchanged after 56 h. No outcome of any kind was assessed.

12. [12]Effect of starvation on human muscle protein metabolism and its response to insulin — Am J Physiol, 1990PMID 2221049

    Design: 8 healthy subjects, forearm amino acid kinetics with labelled phenylalanine and leucine at 12 h and 60 h

    What it does not show: Eight people, forearm muscle, sixty hours. It shows the direction of protein turnover in the first days of a fast; it does not quantify how much muscle anyone loses, and forearm kinetics are not whole-body body composition.

13. [13]Spontaneous 24-h ghrelin secretion pattern in fasting subjects: maintenance of a meal-related pattern — Eur J Endocrinol, 2005PMID 15941923

    Design: 6 healthy volunteers, blood sampled every 20 minutes across a 24 h fast

    What it does not show: Six people, described by its own authors as a pilot. It measures a hunger hormone, not hunger. That the two track each other loosely is an assumption the study does not test.

14. [14]Fasting headache: a review of the literature and new hypotheses — Headache, 2009PMID 19472450

    Design: Narrative review of fasting, Yom Kippur and Ramadan headache literature

    What it does not show: Reviews an association and proposes mechanisms — hypoglycaemia and caffeine withdrawal — without establishing either. It does not tell you how likely a headache is for you, or how to prevent one.

15. [15]Systemic metabolic, hormonal, and glycomic remodeling during a 72-hour fast in healthy adults: a pilot study — Croat Med J, 2026PMID 42286908

    Design: 5 healthy adults, single-arm 72 h water-only fast, no control group

    What it does not show: Five people, single arm, explicitly labelled preliminary by its own authors. They state plainly that autophagy was not measured — only an environment "conducive to" it was inferred.

16. [16]Intermittent fasting activates markers of autophagy in mouse liver, but not muscle from mouse or humans — Nutrition, 2022PMID 35660501

    Design: 50 women on two 8-week intermittent fasting protocols, run alongside a mouse arm

    What it does not show: Does not measure autophagy in human liver, which is where the mouse effect was found, and muscle biopsy markers are not whole-body autophagy. It cannot be used to say fasting fails to induce autophagy in people — only that human muscle markers did not move the way the popular claim predicts.

17. [17]Training state and skeletal muscle autophagy in response to 36 h of fasting — J Appl Physiol (1985), 2018PMID 30161009

    Design: Untrained and trained adults, vastus lateralis biopsies at 2, 12, 24 and 36 h

    What it does not show: Establishes no dose, no threshold hour and no health outcome. The authors’ own summary is that autophagy was "only modestly affected". It provides no basis for the claim that autophagy switches on at a particular hour.

18. [18]Flipping the Metabolic Switch: Understanding and Applying the Health Benefits of Fasting — Obesity (Silver Spring), 2018PMID 29086496

    Design: Narrative review proposing the "metabolic switch" framework

    What it does not show: A framework paper, not a trial. It is the origin of the widely quoted "12 to 36 hours" window, and that range is a synthesis of existing physiology rather than a measurement anyone made.

19. [19]Intermittent Fasting and Obesity-Related Health Outcomes: An Umbrella Review of Meta-analyses of Randomized Clinical Trials — JAMA Netw Open, 2021PMID 34919135

    Design: 11 meta-analyses covering 130 randomised trials; median trial size 38, median follow-up 3 months

    What it does not show: Grades the evidence rather than adding to it, and the grading is the finding: of 104 associations, one was supported by high-quality evidence and 72% by very low quality. It cannot tell you whether fasting will work for you.

20. [20]Effects of Time-Restricted Eating on Weight Loss and Other Metabolic Parameters in Women and Men With Overweight and Obesity: The TREAT Randomized Clinical Trial — JAMA Intern Med, 2020PMID 32986097

    Design: 116 adults with overweight or obesity, 12 weeks, 16:8 vs three structured meals

    What it does not show: Twelve weeks, and neither group was given a calorie target. It tests meal timing alone. It does not test time-restricted eating combined with deliberate calorie restriction, which is how most people actually use it.

21. [21]Calorie Restriction with or without Time-Restricted Eating in Weight Loss — N Engl J Med, 2022PMID 35443107

    Design: 139 adults with obesity, 12 months, calorie restriction with or without an 8 a.m.–4 p.m. window

    What it does not show: Both arms were calorie restricted, so this isolates the timing and nothing else. It does not show that time-restricted eating is useless — it shows that the window adds little once intake is already controlled.

22. [22]Early Time-Restricted Feeding Improves Insulin Sensitivity, Blood Pressure, and Oxidative Stress Even without Weight Loss in Men with Prediabetes — Cell Metab, 2018PMID 29754952

    Design: Supervised controlled-feeding crossover in men with prediabetes, 5 weeks per arm, weight held constant

    What it does not show: Men only, prediabetes only, very small, and called a proof-of-concept study by its own authors. The eating window ended before 3 p.m., which is not the schedule almost anyone follows; the result cannot be transferred to a late window.

23. [23]Diabetes and Ramadan: Practical guidelines 2021 — Diabetes Res Clin Pract, 2022PMID 35016991

    Design: International Diabetes Federation and Diabetes and Ramadan International Alliance guideline

    What it does not show: Written for Ramadan and for clinicians managing it. It is cited here for its risk stratification and for the fact that fasting on insulin or a sulfonylurea requires a prescriber-led plan — not as guidance any reader should apply to themselves unaided.

24. [24]The effect of Ramadan fasting during pregnancy on perinatal outcomes: a systematic review and meta-analysis — BMC Pregnancy Childbirth, 2018PMID 30359228

    Design: 22 studies, 31,374 pregnancies, 18,920 of them exposed to Ramadan fasting

    What it does not show: Observational studies of daylight-hours fasting, not of extended fasting. The review found **no data at all** on perinatal mortality and concluded that evidence on outcomes other than birth weight is insufficient. Absence of a detected effect is not a finding of safety.

25. [25]Engagement in Intermittent Fasting is Prospectively Associated With Higher Body Mass Index, Higher Eating Disorder Psychopathology, and Lower Intuitive Eating in Chinese Adults — Int J Eat Disord, 2025PMID 39530408

    Design: 491 Chinese adults, baseline and 8-month follow-up, cross-lagged analysis

    What it does not show: Observational, so it cannot establish direction of cause — people already prone to disordered eating may be more drawn to fasting. One country, one age band, self-reported measures.

26. [26]Understanding Refeeding Syndrome in Critically Ill Patients: A Narrative Review — Nutrients, 2025PMID 40507135

    Design: Narrative review of refeeding syndrome in intensive care

    What it does not show: Concerns critically ill hospital patients. It is cited for the mechanism of the electrolyte shift that follows refeeding, and explicitly **not** to suggest an ordinary fast in a well-nourished adult carries comparable risk.

27. [27]Influence of glucagon on natriuresis and glucose-induced sodium retention in the fasting obese subject — Eur J Clin Invest, 1977PMID 408142

    Design: 37 non-diabetic subjects with obesity, 7-day supervised fast with infusion arms

    What it does not show: A supervised fast in people with obesity, studying the mechanism of sodium loss. It establishes that fasting causes real urinary sodium loss, peaking early; it does not tell you how much salt to replace, and nothing on this page does.

28. [28]Influence of insulin and glucagon on sodium balance in obese subjects during fasting and refeeding — Int J Obes, 1981PMID 6113218

    Design: Review of sodium handling across fasting and refeeding in subjects with obesity

    What it does not show: Same population and the same limits as the study above. Cited for the direction of the effect across fasting and refeeding, not for any dose.

29. [29]Exercise-associated hyponatremia — Clin J Am Soc Nephrol, 2007PMID 17699400

    Design: Clinical review of dilutional hyponatraemia in endurance athletes

    What it does not show: Concerns athletes drinking heavily during exercise, not fasting. The mechanism — blood sodium diluted by a large plain-water intake — is the same one that makes drinking litres of water to suppress hunger during a long fast a genuinely dangerous idea.

30. [30]Diagnosis and Management of Sodium Disorders: Hyponatremia and Hypernatremia — Am Fam Physician, 2023PMID 37983699

    Design: Clinical review for physicians

    What it does not show: Not about fasting. Cited only for what low blood sodium does and how it presents, so that the warning above it is checkable rather than assertive.