Who Should Not Try Intermittent Fasting

Fasting is rarely the hazard by itself — it becomes one in combination#

For most healthy adults a daily eating window is uneventful. The groups who should not run one are not defined by fragility in general but by a specific interaction: fasting plus a medication that lowers blood glucose, fasting plus a training load already outstripping intake, or fasting plus a history that makes food rules dangerous. In each case the fasting schedule is the second ingredient, not the poison — which is why the standard warning list is mostly correct and almost never argued properly.

And one entry on that list has an actual number behind it while the others do not. In a randomized trial in people with type 2 diabetes, two very-low-calorie days a week roughly doubled the rate of hypoglycemia even after clinicians had reduced medication in advance. That is a quantified, mechanistic risk. Compare it with pregnancy, where the caution is universal in practice and the evidence is much thinner and more surprising than the confidence around it. The pillar covers what fasting does for weight; this page is about the four situations where the answer changes, and how much is actually known about each.

Glucose-lowering medication: the one risk with a number on it#

Insulin and sulfonylureas lower blood sugar on a schedule set by the dose, not by whether you ate. Remove the meal and keep the dose and the arithmetic runs one way. That much is obvious. What is less obvious is how much of the risk survives careful management.

Forty-one adults with type 2 diabetes — BMI 30 to 45, on metformin and/or glucose-lowering medication, HbA1c 50 to 86 mmol/mol — followed a roughly 500-calorie diet on two days a week for 12 weeks, randomized to consecutive or non-consecutive fast days. Participants received education and had their medication reduced in advance. Fasting still increased hypoglycemia relative to non-fasting days, with a rate ratio of 2.05 (95% CI 1.17 to 3.52), and it made no significant difference whether the two fast days were consecutive or spread apart (RR 1.54, 95% CI 0.35 to 6.11)¹.

Read both halves of that result. The risk was real and roughly doubled despite pre-emptive dose reduction — so the informal advice to "just skip a dose on fast days" is not a solution someone should improvise. But the absolute rate was low, averaging 1.4 events over 12 weeks, and both arms improved on weight, HbA1c, fasting glucose and quality of life. This is not a group who cannot benefit from fasting. It is a group for whom the medication adjustment is the intervention and the diet is the easy part. A separate year-long trial found glycemic events affecting a third of participants on sulfonylureas or insulin during the first fortnight — the numbers are in the 5:2 diet explained. Either way, the sequence is medication review first, fast day second, and never the reverse.

Athletes: the problem is energy availability, and the window makes it worse by accident#

The athletic caution is often stated as "fasting hurts performance," which is both vague and mostly untrue for a single session. The real construct is energy availability — energy intake set against the energy your training burns — and the syndrome that follows when it stays too low.

The International Olympic Committee's consensus statement defines Relative Energy Deficiency in Sport as a syndrome of deleterious health and performance outcomes in female and male athletes exposed to low energy availability, meaning inadequate energy intake relative to exercise energy expenditure². Note what the definition does not contain: a schedule. Nothing in it says an eating window is harmful. The risk is arithmetical.

Which is exactly why a window is a poor tool for someone training hard. Compressing an athlete's food into eight hours does not lower their expenditure; it lowers the number of opportunities to meet it. Two or three eating occasions against a 3,500-calorie requirement is a structurally harder problem than five, and the failure is silent — nobody notices the day they came in 400 calories short. That is the mechanism by which a schedule adopted for convenience becomes a chronic shortfall, and it is a different failure from the muscle-loss question, where protein and training are the deciding variables. If you are training at volume and your intake is already effortful to hit, adding a constraint on when you may eat is solving the wrong problem.

Disordered eating: the evidence just got harder to dismiss#

The caution here used to rest entirely on cross-sectional surveys, which can only show that people who fast score higher on eating-disorder measures — never which came first. That layer exists: among 64 adults practising roughly 16-hour daily fasts, men and women alike scored significantly above community norms on every subscale of the Eating Disorder Examination Questionnaire, and 31.25% scored at or above the clinical cut-off³. A reasonable person could read that and conclude the causation runs backwards: people already prone to restriction find fasting appealing.

A longitudinal study makes that reading harder to hold. Four hundred and ninety-one Chinese adults, mean age 30, were assessed at baseline and again eight months later. Relative to never engaging in intermittent fasting, both current and past engagement predicted higher eating-disorder psychopathology and greater eating-related psychosocial impairment at follow-up, and past engagement predicted higher BMI and lower unconditional permission to eat⁴. The authors' framing is that intermittent fasting may carry the same enduring effects on eating behavior as traditional dieting.

The finding that should give a fasting enthusiast pause is not the current fasters. It is the past fasters, who had stopped and still scored worse eight months on.

Hold the limits in view: 491 people, one country, self-report, an eight-month window, and an observational design that cannot rule out a shared underlying cause. This is not proof that fasting causes eating disorders. It is enough to say the direction-of-causation defense no longer does all the work it used to, and that anyone with a history of restriction, bingeing or compensatory behaviour is choosing a tool whose central mechanic is a rule about when eating is permitted. For that person the difference between structure and rigidity matters more than any protocol detail, and a less rule-shaped approach is the safer starting point.

Pregnancy: the caution is near-universal and the evidence is thinner than you'd guess#

Every list says do not fast while pregnant, and that remains the sensible default. But the largest body of evidence on the question does not say what most people assume, and pretending otherwise is its own kind of carelessness.

A systematic review and meta-analysis pooled 22 studies covering 31,374 pregnancies, 18,920 of them exposed to Ramadan fasting. Birth weight was not affected (standardised mean difference 0.03, 95% CI 0.00 to 0.05), and preterm birth was not increased (odds ratio 0.99, 95% CI 0.72 to 1.37). Placental weight was significantly lower in fasting mothers, though that result was dominated by a single large study. The reviewers' conclusion was that Ramadan fasting does not adversely affect birth weight, with insufficient evidence regarding other perinatal outcomes⁵.

Three things stop that from being a green light, and they are worth stating precisely rather than waving away. Ramadan fasting is dawn-to-dusk with an eating night attached — closer to a shifted eating window at energy balance than to a deliberate calorie deficit. The review found no usable data at all on perinatal mortality or small-for-gestational-age infants, which are the outcomes that would matter most. And the whole literature is observational: women who feel unwell tend to stop fasting, which quietly removes the highest-risk pregnancies from the exposed group. So the accurate position is that the population most studied shows no birth-weight harm from a Ramadan-style fast, and that this tells you very little about running a weight-loss window during pregnancy — which nobody has tested, and which is a change to make with your clinician rather than from an article.

The list, and what each entry actually rests on#

Group	The actual mechanism	Strength of evidence
On insulin or a sulfonylurea	Dose keeps working when the meal doesn't arrive	Strong — RR 2.05 for hypoglycemia despite dose reduction¹
Training at volume	Fewer eating occasions against an unchanged expenditure	Strong mechanism, consensus-level²
History of disordered eating	A rule about permitted eating hours	Observational, now with prospective support⁴
Pregnant or breastfeeding	Growth demand plus untested deficit	Weak and indirect — no data on the outcomes that matter⁵
Underweight, or still growing	No margin for an accidental shortfall	Reasoned, not trialled

If you are on this list, the useful move is not to find a gentler fasting protocol. It is to change the variable the protocol was a proxy for. Fasting appeals because it removes eating occasions without asking you to think about amounts — so for anyone who can't safely remove eating occasions, the substitute is low-effort awareness of the amounts themselves, which costs nothing metabolically and is compatible with medication schedules, training loads and pregnancy alike. Paired with a rate of loss that isn't aggressive, that gets you the same deficit through a door that isn't locked. And if none of these apply to you, the ordinary on-ramp is in how to start intermittent fasting; the sex-specific questions get their own treatment in intermittent fasting for women.

FAQ#

Why does fasting cause low blood sugar even when the medication dose is reduced?#

Because insulin and sulfonylureas lower glucose on a fixed schedule regardless of intake, and the size of the pre-emptive dose cut is a guess. In a 12-week randomized trial where participants were educated and had medication reduced in advance, fast days still carried roughly double the rate of hypoglycemia (RR 2.05, 95% CI 1.17 to 3.52)¹. Absolute rates were low, but the adjustment is a clinical one, not a self-managed one.

Is intermittent fasting safe during pregnancy?#

The default answer is no, and it should stay no — but the reason is missing evidence rather than demonstrated harm. Across 31,374 pregnancies, Ramadan fasting did not affect birth weight or preterm birth, while producing no usable data on perinatal mortality or small-for-gestational-age infants⁵. A Ramadan fast is also not a weight-loss window: it shifts eating hours at roughly energy balance rather than creating a deliberate deficit, and nobody has tested the latter in pregnancy.

Does intermittent fasting cause eating disorders?#

Not demonstrably — but the evidence has moved past pure correlation. Cross-sectional work found 31.25% of intermittent fasters scoring at or above the clinical cut-off on the EDE-Q³, and an eight-month longitudinal study found both current and former fasters showing higher eating-disorder psychopathology at follow-up⁴. That persistence after stopping is what makes the "they were already like that" explanation insufficient on its own.

Sources#

This article was researched and drafted with AI assistance and reviewed for accuracy by the BurnWeek team. It is general information, not medical advice. How we research and correct our articles →