Does intermittent fasting actually work for weight loss? How the control group changes the answer, the blank after 24 weeks, and where three sources disagree, 2026

TL;DR Intermittent fasting is not one way of eating. It is at least three. The American Diabetes Association's Standards of Care in Diabetes for 2026 divides it into alternate-day fasting, the 5:2 pattern, and time-restricted eating; the "168" people talk about is the shortest window within time-restricted eating. The evidence for the three is not interchangeable. "Compared with what" decides the answer — the studies are not contradicting each other. Against "carry on eating as before", a 2025 Nutrition Journal meta-analysis (15 randomised controlled trials, 758 people) reports an average loss of 3.73 kg. Against "ordinary dietary advice", the Cochrane systematic review of 2026 (21 trials, 1,430 people on that comparison) reports a difference in percentage of baseline weight lost of MD −0.33 percentage points (95% CI −0.92 to 0.26). Against "the same calorie deficit", the 2025 BMJ network meta-analysis (99 trials, 6,582 people) finds −1.29 kg (95% CI −1.99 to −0.59) for alternate-day fasting alone. Three different questions, three consistent answers. Duration is the real watershed. BMJ 2025 stratified by trial length: trials of 24 weeks or more (n=17) showed a weight effect only against ad-libitum eating — and the authors themselves state that this "not seen" may reflect an insufficient number of long enough studies rather than a demonstrated absence. The BMJ authors' own minimally important difference is 2.0 kg, and they say plainly that the −1.29 kg for alternate-day fasting against continuous energy restriction falls below that level. BMJ 2025 identified only five trials of 52 weeks or longer. "Intermittent fasting improves your lipids" does not hold up. The same Nutrition Journal analysis reports triglycerides rising by 13.22 mg/dl at 12 weeks or less; Trepanowski's 12-month trial in 2017 reports LDL cholesterol in the alternate-day fasting group 11.5 mg/dL higher than in the continuous calorie restriction group (with no difference between the groups at month 6 in that same trial, whose participants were mainly metabolically healthy obese adults — which is how the authors themselves explain why most cardiovascular markers did not move). This article will not tell you to fast, and will not tell you not to, and it will certainly not advise on any drug dose. If you have diabetes, are pregnant, are taking insulin or a sulfonylurea, or live with any long-term condition, whether fasting suits you and how your medication should be adjusted is a decision only your doctor or diabetes team can make.

#Which way of eating does "intermittent fasting" actually mean?

The term covers at least three completely different arrangements, and any given study has tested only one of them — so a headline saying intermittent fasting does or does not work is quite likely talking about the one you are not doing.

The most authoritative classification available comes from Chapter 5 of the ADA's Standards of Care in Diabetes—2026: intermittent fasting is an umbrella term covering three main forms — alternate-day fasting (energy restriction to 500–600 calories on alternate days), the 5:2 eating pattern (energy restriction to 500–600 calories on two days, consecutive or not, with usual intake on the other five days), and time-restricted eating (daily calorie restriction achieved through an eating window of 8 to 15 hours) [Note 1].

In one sentence: the ADA's "time-restricted eating" is an eating window of 8 to 15 hours, and the "168" people talk about (8 hours of eating, 16 hours not) is the shortest end of that range.

The awkward part is that the three main documents use three different classifications.

Cochrane 2026's inclusion criteria: randomised controlled trials and cluster-RCTs comparing intermittent fasting (including time-restricted feeding, periodic fasting, alternate-day fasting and modified alternate-day fasting) with regular dietary advice, no intervention or a waiting list, in men and women with overweight or obesity, with or without associated comorbid conditions; the minimum duration of the intervention was four weeks and the minimum duration of follow-up six months; cross-over and quasi-RCTs were excluded [Note 2].

Keep that "minimum follow-up of six months" in mind: it means this review does not include a great many short trials, so its −0.33 percentage points is not a "three-month result". When this article calls Cochrane short-term evidence, the accurate range is six to twelve months, not shorter.

Even "alternate-day fasting" has more than one definition. The protocol in Trepanowski's 2017 trial was 25% of energy needs on fast days and 125% on feast days; the ADA writes 500–600 calories on alternate days; Varady's 2022 review (Nat Rev Endocrinol 2022;18(5):309-321) writes 0–500 calories on fast days with ad-libitum eating on feed days (that review's own wording is "the most studied forms", describing the state of the literature rather than laying down a definition). These are not the same thing.

⚠ A declaration of interests, which this article has already made for another document and so must make here too: the competing-interests statement for Varady 2022 records that the corresponding author K.A.V. received an author fee from a publisher for a book on alternate-day fasting — the same author and the same book as the one listed for Trepanowski 2017 below. This article cites the review only for a wording of the definition, but having disclosed for one document it should not stay silent for another.

⚠ If a headline saying "research proves fasting works" or "research proves it doesn't" makes you want to follow it: before you start, ask three questions — which form did it test? for how long? and against what? If those three cannot be answered, the number means nothing for you.

#Why do you see "3.7 kg lost" and "almost no difference" at the same time?

Because they are not answering the same question. The three documents ask, respectively, whether fasting beats carrying on as before, whether it beats the dietary advice a doctor would have given anyway, and whether it beats the same calorie deficit — and the three answers are entirely consistent with each other.

The ADA says the same, and adds something the table above does not have. The "Nonreligious Fasting" section of ADA Chapter 5 states that each of the three forms produces mild to moderate weight loss (a 3% to 8% loss from baseline) over short durations (8 to 12 weeks), with no significant difference in weight loss compared with continuous calorie restriction [Note 3].

Notice two things. First, "8 to 12 weeks" is the timescale the ADA itself attaches — this is a short-term figure, not a figure at one year. Second, this is not the same conclusion as the third row of the table above, and the two must be kept apart. The ADA sentence combines the three forms and concludes "no significant difference" against continuous calorie restriction; but BMJ 2025, cited in the third row, does not answer "no difference" — the BMJ found alternate-day fasting significantly beating continuous calorie restriction (−1.29 kg, moderate certainty). The two sources disagree on that one point, alternate-day fasting versus continuous energy restriction. This article reports both and will not present them as one conclusion.

⚠ And a 2024 pooled analysis the ADA cites says something worth stopping on: the authors concluded that intermittent energy restriction produces small but statistically significant reductions in waist circumference and fat-free mass, but was otherwise not superior to continuous energy restriction eating patterns [Note 4].

The words to watch there are "fat-free mass" — muscle and other non-fat tissue. In weight-loss research a falling waist is normally counted as good, but falling fat-free mass is not: it means that what came off was not all fat. That sentence does not contradict the table, but it is about more than how much came off — it is about what came off, and that is a column most popular articles leave out.

#This is not "conflicting evidence", it is a division of labour by design

The Nutrition Journal meta-analysis of 2025 — by Wang B, Wang C and Li H, published 30 July 2025, PROSPERO registration CRD420251036588, literature searched to March 2025 — explicitly excludes studies with a calorie-restriction control group. Its inclusion criteria state that the control group means maintaining habitual dietary patterns, with exercise groups also treated as controls; that studies comparing against a calorie-restriction group in participants who are not overweight or obese are listed as exclusions; and that where a study has an intermittent fasting group, a calorie restriction group and a control group, only the comparison between the fasting group and the control group is included. (That is a restatement of the inclusion criteria, not a verbatim quotation.)

Which is to say: the −3.73 kg figure could not, by design, be "fasting beats the same calorie deficit". It answers the difference between not changing your eating and starting to fast.

Cochrane 2026 answers a different question: compared to regular dietary advice, intermittent fasting may result in little to no difference in percentage from baseline weight loss (MD −0.33, 95% CI −0.92 to 0.26; 21 studies, 1,430 participants; low-certainty evidence due to risk of bias) [Note 5]. BMJ 2025 answers a third: all intermittent fasting and continuous energy restriction strategies reduced body weight compared with an ad-libitum diet; and compared with continuous energy restriction, alternate-day fasting was the only form to show a benefit in body weight reduction (−1.29 kg, 95% CI −1.99 to −0.59, moderate certainty) [Note 6].

Do the arithmetic: Cochrane's result is in percentage points of baseline weight, not kilograms. Take someone of 70 kg: MD −0.33 percentage points is 70 times 0.33%, about 0.23 kg; the two ends of the confidence interval, −0.92 to 0.26 percentage points, convert to a range from about 0.64 kg more lost to about 0.18 kg less lost. In other words, on the question of "against regular dietary advice", the gap between the two is too small to read on a bathroom scale.

And the same Cochrane review's other arm, against no intervention or a waiting list, gives MD −3.42 percentage points (95% CI −4.95 to −1.90; six studies, 427 people, moderate certainty) — about 2.4 kg for a 70 kg person, with a range of roughly 1.3 to 3.5 kg. That points the same way as the −3.73 kg.

#⚠ Two gaps the reader is owed

First, the wording of the Cochrane abstract does not quite match its own figures. The sentence on the "no intervention" arm reads "likely results in little to no difference", but the estimate printed in the same sentence is MD −3.42 (95% CI −4.95 to −1.90), moderate certainty — a confidence interval that does not cross zero. This article will not decide for Cochrane which formulation counts; it sets the two side by side and takes the "regular dietary advice" arm, where the wording and the figures agree, as its principal citation.

Second, the Cochrane press release and the review itself do not agree. The press release (Parkinson M, Cochrane, 16 February 2026) says that intermittent fasting is unlikely to lead to greater weight loss in overweight or obese adults than traditional dietary advice or doing nothing [Note 7]. The release compresses two control arms into one sentence, but the review itself prints −3.42 percentage points on the "no intervention" arm. Wherever a weight figure is involved, this article cites the review and not the press release.

⚠ If contradictory health headlines have ever made you think the experts have no consensus: this lesson is not confined to fasting — whenever you see an effect figure for any treatment or diet, find the control group first. Change the control group and the same intervention can go from "significantly effective" to "no difference", and both conclusions can be true.

#And after six months, or a year?

This is the most important finding: the advantage of intermittent fasting over the same calorie deficit is not seen in trials of 24 weeks or more — and the BMJ authors state themselves that this may reflect an insufficient number of long enough studies rather than a demonstrated absence; within their search they identified only five trials of 52 weeks or longer.

BMJ 2025 stratified by follow-up: estimates were similar among trials with less than 24 weeks of follow-up (n=76), but moderate-to-long-term trials (24 weeks or more, n=17) showed benefits in weight reduction only when the diet strategies were compared with ad-libitum eating [Note 8]. The results section of the full paper is more specific: in trials of 24 weeks or more, the most restrictive strategies (alternate-day fasting, time-restricted eating and continuous energy restriction) showed small body weight reductions versus ad-libitum (mean difference range −1.88 to −3.63 kg), with no differences between the intermittent fasting strategies and continuous energy restriction in these moderate-to-long-term studies [Note 8].

#⚠ But the authors added a sentence of their own, and it changes how the paragraph above reads

The BMJ discussion says immediately afterwards: the loss of association in the network assessment of moderate-to-longer-term trials (24 weeks or more) may be due to an insufficient number of studies available [Note 9].

In one sentence: "the advantage disappears after 24 weeks" and "there are too few trials past 24 weeks to see it" cannot be separated within this analysis, and the authors say so themselves. 17 trials have to be spread across a network of four diet strategies. So when this article says the advantage disappears, the accurate formulation is: it is not seen in trials of 24 weeks or more, and the authors attribute that "not seen" to an insufficient number of studies — "not seen" is not "shown absent", just as "seen" is not "clinically meaningful".

#And the −1.29 kg, the authors say, is not big enough either

The threshold the BMJ authors set for themselves is stated in the legend to figure 4: the minimally important difference for body weight was 2.0 kg [Note 10]. They compare their own result against that line in the discussion: while many of the reductions observed with intermittent fasting strategies compared with ad-libitum exceeded this threshold, individual strategies — particularly alternate-day fasting versus continuous energy restriction at −1.29 kg — fell below this level, suggesting somewhat limited clinical impact in those comparisons [Note 10].

Do the arithmetic: 2.0 minus 1.29 is 0.71 kg. Which is to say that the one statistically significant advantage of alternate-day fasting over the same calorie deficit falls 0.71 kg short of the line the authors themselves drew for clinical meaning, reaching only 64.5% of the threshold (1.29 divided by 2.0).

The authors' summary sentence: the results show that intermittent fasting may offer unique benefits primarily in the short term, whereas both intermittent fasting and continuous energy restriction appear to provide similar moderate-to-long-term improvements over ad-libitum diets, and this equivalence in sustained outcomes is a critical take-away for clinicians managing chronic metabolic conditions [Note 11].

#The one-year single trials say the same thing

Liu D et al., N Engl J Med 2022;386(16):1495-1504 (Nanfang Hospital, Guangzhou; NCT03745612): 139 patients with obesity randomised to "time-restricted eating (8am to 4pm) plus calorie restriction" or "daily calorie restriction alone", both groups prescribed 1,500–1,800 calories a day for men and 1,200–1,500 for women, for 12 months. Weight change at 12 months: −8.0 kg in the time-restricted group (95% CI −9.6 to −6.4) against −6.3 kg in the calorie-restriction-only group (95% CI −7.8 to −4.7), a net between-group difference of −1.8 kg (95% CI −4.0 to 0.4, P = 0.11). The original conclusion is at [Note 12]: among patients with obesity, a regimen of time-restricted eating was not more beneficial with regard to reduction in body weight, body fat or metabolic risk factors than daily calorie restriction. Do the arithmetic: 118 of the 139 (84.9%) completed 12 months of follow-up, so 21 were lost, 15.1%.

Trepanowski JF et al., JAMA Intern Med 2017;177(7):930-938 (Chicago; NCT00960505): 100 people randomised to alternate-day fasting (25% of energy needs on fast days, 125% on feast days), daily calorie restriction (75% every day) or a no-intervention control, for one year (six months of weight loss followed by six months of maintenance). Weight change against control at month 6: alternate-day fasting −6.8% (95% CI −9.1% to −4.5%) against calorie restriction −6.8% (95% CI −9.1% to −4.6%); at month 12: −6.0% (95% CI −8.5% to −3.6%) against −5.3% (95% CI −7.6% to −3.0%). The original conclusion is at [Note 13]: alternate-day fasting did not produce superior adherence, weight loss, weight maintenance or cardioprotection versus daily calorie restriction.

#Adherence: the fasting group did not fail to want to, they failed to manage it

Dropout in that same trial: of the 100 participants, the alternate-day fasting group had the highest dropout, 13 of 34 (38%); the daily calorie restriction group 10 of 35 (29%); the control group 8 of 31 (26%) [Note 14]. And what they did: participants in the alternate-day fasting group ate more than prescribed on fast days and less than prescribed on feast days, while those in the daily calorie restriction group generally met their prescribed energy goals [Note 14].

In other words, in this trial the fasting group's actual behaviour drifted towards the middle — the fast days were not as sparse and the feast days not as generous — and their dropout was the highest of the three.

⚠ But the authors state in their own limitations that this dropout rate may itself have contaminated the result: because the dropout rate was higher than anticipated, their power to detect a 5% difference in weight loss between the intervention groups at month 6 fell from 80% to 60%; and the higher dropout in the alternate-day fasting group may also have introduced a selection bias between groups [Note 15].

Which is to say that the 38% carries two meanings, and they point in opposite directions. On one hand it is a real signal about whether people can keep it up; on the other, it leaves behind a group of people unusually able to keep it up, which weakens the conclusion that alternate-day fasting is not superior to daily calorie restriction. Both points are reported here; the half that suits the argument is not taken on its own.

And this is not a one-trial phenomenon. BMJ 2025, immediately after its account of adverse events, sets out adherence across all 99 trials, and separates short-term from long-term: of the 99 studies, 74 included a description of adherence or enough information to determine it; most trials reported high adherence (above 80%) throughout; and among the trials with a 52-week follow-up, most stated poor adherence in the intervention groups — in one study, adherence to whole day fasting declined from 74% at six weeks to 22% at 52 weeks; by contrast, however, another 52-week study reported high adherence in both the time-restricted eating and continuous energy restriction groups, both approximately 84% [Note 16].

In one sentence: in the short term most people manage it, and in the one-year trials most do not; yet in another one-year trial both groups managed about 85%. Results in both directions exist, the BMJ printed both, and this article reports both — this box cannot be filled with a clean answer, and it would be the filling-in that ought to worry you.

#And how thin is the long-term evidence?

BMJ 2025's own stated limitation: the scarcity of trials of 52 weeks or longer — only five identified — precluded a separate network meta-analysis for very long-term effects, limiting relevance to sustained long-term weight loss [Note 17]. Cochrane 2026 says the same thing: the included studies focused on short-term effects of the intervention (up to 12 months), limiting the applicability of the evidence in the review to decision-making over longer durations [Note 17].

⚠ If you are planning to "try it for three months and see": three months falls in the BMJ's "less than 24 weeks" stratum — the one in which a difference is easiest to see. The advantage seen at that stage is not seen in trials past 24 weeks — and since the authors say that may be down to an insufficient number of studies, the correct conclusion is not "you will regain it after three months" but this: what happens after three months is a question no one yet has enough long enough trials to answer for you.

#Beyond weight, what about the other markers?

"Intermittent fasting improves your lipids" does not hold up — the same body of literature reports both triglycerides rising and LDL cholesterol rising, while the glycaemic markers show no difference at all against the same calorie deficit.

#Blood glucose and lipids: against the same calorie deficit, almost all no difference

BMJ 2025 states plainly: no differences were noted between intermittent fasting, continuous energy restriction and ad-libitum diets for glycated haemoglobin and high density lipoprotein; and alternate-day fasting was the only strategy to show an improvement in anthropometric and lipid measures compared with continuous energy restriction, with no benefit observed for glycated haemoglobin or high density lipoprotein cholesterol in any diet strategy comparison [Note 18]. And comparing strategies against each other, the BMJ reports that time-restricted eating raised total cholesterol, LDL cholesterol and non-HDL cholesterol relative to whole day fasting [Note 18].

#Three concrete figures where the lipids got worse

One — Trepanowski 2017 (12 months, alternate-day fasting against daily calorie restriction): by month 12, mean LDL cholesterol levels were significantly elevated in the alternate-day fasting group, by 11.5 mg/dL (95% CI 1.9 to 21.1), compared with the daily calorie restriction group [Note 19].

⚠ Two sentences immediately before that one have to be read with it: first, LDL cholesterol showed no significant difference between the groups at month 6 — the +11.5 mg/dL appears only at month 12, it was not high throughout. Second, total cholesterol showed no significant difference between the two intervention groups, or against the control, at either month 6 or month 12. The same trial also reports HDL cholesterol in the alternate-day fasting group significantly raised by 6.2 mg/dL against the calorie restriction group at month 6 (95% CI 0.1 to 12.4), but no longer significant at month 12 (1.0 mg/dL, 95% CI −5.9 to 7.8). Blood pressure, heart rate, triglycerides, fasting glucose, fasting insulin, insulin resistance, C-reactive protein and homocysteine showed no significant difference between the groups at either month 6 or month 12.

⚠ And the authors explain in the discussion why most of the cardiovascular markers did not move, which changes the weight of the whole set of figures: the trial included primarily metabolically healthy obese adults, and since many participants had normal cholesterol levels and normal blood pressure at baseline, it is not surprising that most cardiovascular risk indicators did not change in response to diet [Note 20]. In one sentence: these participants' lipids and blood pressure were largely normal to begin with, so "no improvement" was, in the authors' own view, to be expected, and cannot be taken as a general rule that fasting does nothing for lipids. The raised LDL figure still stands, but it was measured in a metabolically healthy group.

Two — Nutrition Journal 2025 (against eating as usual): the analysis reports that short-term intermittent fasting of 12 weeks or less may temporarily raise triglycerides, MD 13.22 mg/dl (95% CI 3.39 to 23.05), while interventions longer than 12 weeks optimise the lipid effect; total cholesterol MD −6.31 mg/dl (95% CI −12.36 to −0.26); diastolic blood pressure MD −3.30 mmHg (95% CI −5.47 to −1.13); with no significant effect on systolic blood pressure, fasting glucose or glycated haemoglobin. The authors' own stated limitations are that most studies had short intervention periods (12 weeks or less) and high heterogeneity, and that standardised long-term trials are needed to verify sustained benefit and safety.

Three — fat-free mass: Schroor MM et al. (Adv Nutr 2024;15(1):100130) pooled 28 trials covering 2 to 52 weeks of time-restricted eating (7), alternate-day fasting (10) and 5:2 (11), against continuous energy restriction: all intermittent energy restriction diets combined reduced fat-free mass (WMD −0.20 kg, 95% CI −0.39 to −0.01) and waist circumference (−0.91 cm, 95% CI −1.76 to −0.06) more than continuous energy restriction; effects on body mass index, glucose, insulin, the insulin resistance index, serum lipid and lipoprotein concentrations and blood pressure did not differ; further, time-restricted eating reduced body weight, fat mass and fat-free mass more than continuous energy restriction, whereas alternate-day fasting improved the insulin resistance index more; and BMI was reduced less on the 5:2 diet than on continuous energy restriction [Note 21].

Those last two clauses have to be read together, because they point in opposite directions: alternate-day fasting wins on insulin resistance, and 5:2 loses on BMI. This analysis is not a blanket verdict against fasting — each strategy and each marker comes out differently, which is why "is fasting good or bad" is too coarse a question to ask.

⚠ The "conclusion" sentence of that same analysis is in tension with the sentence immediately after it, and both must be read. Schroor's authors conclude that the three intermittent energy restriction diets combined did not produce superior improvements in anthropometric and cardiometabolic risk markers; but immediately afterwards they themselves add that slightly greater reductions were, however, observed in fat-free mass and waist circumference [Note 21]. Which is to say that Schroor's own authors acknowledge fat-free mass and waist circumference as exceptions, not to be swallowed by the "not superior" summary. The authors separately add that energy intake may not have been equal between study groups, which may have affected these results.

In other words, the popular claim that fasting costs you muscle corresponds, at the pooled level, to an actual figure of −0.20 kg of fat-free mass, with the upper end of the confidence interval close to zero. That is a small but consistently directed difference — not a large loss, and not nothing either.

⚠ If you are fasting for your lipids or your blood glucose rather than for your weight: at the same calorie deficit, no benefit was observed for glycated haemoglobin or HDL cholesterol in any comparison, and LDL moved upwards in the one-year alternate-day fasting trial. If your target is a lipid or glucose number, that is something to decide with your doctor, not something an eating timetable handles on its own.

#What are the side effects? How far does the evidence go?

What the trials report is mostly mild discomfort, appearing early and settling afterwards; but the reporting of adverse events is itself inconsistent and of very low certainty — which is to say that existing trials cannot rule out harm, which is not the same as harm having been ruled out.

BMJ 2025 on adverse events: 56 trials assessed adverse events, of which 27 reported no harmful events in the fasting groups; most trials reported mild side effects, including constipation, nausea, hunger, diarrhoea and dizziness [Note 22]. The same passage also records one serious event by name: Holmer et al. reported a participant who became hypoglycaemic and fell, and who remained in the study; another study reported a case of high bilirubin and low blood sodium, and a participant with low potassium at the end of a 26-week trial.

Cochrane 2026's adverse-event estimates: against regular dietary advice, RR 1.45 (95% CI 0.64 to 3.28; 7 studies, 619 people, very low certainty); against no intervention or a waiting list, RR 1.84 (95% CI 0.88 to 3.85; 2 studies, 189 people, very low certainty). The Cochrane press release says as much: reporting of side effects was inconsistent across trials, making it difficult to draw firm conclusions, and the evidence base remains limited, with only 22 trials, many with small sample sizes and inconsistent reporting [Note 23].

A review on the British Dietetic Association's website signed by Titmus M and published 28 September 2021 (note: this is a signed commentary by a practising dietitian, not a BDA position statement, and it predates every systematic review cited here) states: no serious adverse implications of time-restricted eating were reported in the literature; mild events such as headaches, diarrhoea, increased thirst and dizziness were potentially related to the intervention, but occurred in the early stages and resolved by week three following adjustment [Note 24].

#One signal at the behavioural level

The same 2021 review also records that time-restricted eating may inadvertently exacerbate existing vulnerability to overeating [Note 25]. It describes participants moving foods such as sweets they would normally eat in the evening into the daytime, and others eating excessively at meals to pre-empt hunger they expected during the fast.

⚠ But the sentence immediately after changes the nature of that signal: of note is that participants ultimately realised that they did not experience excessive hunger, and discontinued the deliberate overeating behaviour [Note 25]. Which is to say that this was a transitional behaviour that resolved by itself, not a persisting harm. The source of those descriptions is a qualitative interview study by Bjerre et al., 17 participants drawn from a randomised controlled trial of 12 weeks with a ten-hour eating window — interview records, not pooled data. The same section also states that individuals considering time-restricted eating will need to be informed of potential transient effects to support adherence.

#The claim going round that "168 raises cardiovascular death risk by 91%" — where does it stand now?

Two things have to be said clearly: first, the retort that "the study was never peer reviewed" is, as of August 2026, no longer true; and second, the "91%" is a preliminary figure that has been superseded. The published hazard ratio is 2.35, and it comes from an observational study.

#What happened in 2024

The American Heart Association newsroom released an item on 18 March 2024 about abstract P192 at the EPI|Lifestyle Scientific Sessions 2024 (senior author: Victor Wenze Zhong of Shanghai Jiao Tong University School of Medicine), reporting that people who ate all of their food within less than 8 hours a day had a 91% higher risk of death from cardiovascular disease [Note 26].

The AHA marked the release itself: abstracts presented at its scientific meetings are not peer reviewed but are curated by independent review panels, and findings are considered preliminary until published as a full manuscript in a peer-reviewed journal [Note 26]. The senior author states in the same release that although the study identified an association between an 8-hour eating window and cardiovascular death, this does not mean that time-restricted eating caused cardiovascular death [Note 26].

#It was then published, and the figure changed

The same NHANES data and the same senior author have now appeared as a formal paper: Chen M, Xu L, Van Horn L et al., Diabetes Metab Syndr 2025;19(7):103278 (PMID 40849219). The study included adult participants of the National Health and Nutrition Examination Survey from 2003 to 2018 who reported usual intake from two valid 24-h dietary recalls, 19,831 people in all; over a median follow-up of 8.1 years, compared with an eating duration of 12 to 14 hours, an eating duration of less than 8 hours was robustly associated with higher cardiovascular mortality (HR 2.35, 95% CI 1.39–3.98), but not with all-cause or cancer mortality; the positive association with cardiovascular mortality remained consistent across 8 subgroups stratified by race and ethnicity, socioeconomic factors and smoking status, and survived 14 sensitivity analyses; the association with all-cause mortality, however, did not survive many sensitivity analyses [Note 27].

The middle clause deserves particular attention, because it runs the opposite way to the fourth point below. The authors' "robustly associated" is not idly chosen — it refers to the cardiovascular mortality association surviving 8 subgroups and 14 sensitivity analyses. What did not hold up was the all-cause one. The two results must not be run together: the cardiovascular signal is the one that stands.

The authors' own conclusion: although a positive association was observed between an eating duration of less than 8 hours and cardiovascular mortality, further research is required to understand whether this risk is attributable to the short eating duration itself or to residual confounding resulting from its contributing factors [Note 28].

#In one sentence

The "not peer reviewed" retort no longer works — the paper has been published. The "91%" should no longer be quoted — what was published is a hazard ratio of 2.35 (about 135% higher), with a confidence interval of 1.39 to 3.98, against a comparison group whose daily eating duration was 12 to 14 hours. But it is an observational study, not a trial.

The AHA published an independent comment alongside it by Dr Christopher D. Gardner (Stanford University, chair of the writing committee for the AHA's 2023 scientific statement Popular Dietary Patterns). His comment has four points, and all four are set out here, because taking only two of them would read as though he were merely raising doubts. His opening sentence is: overall, this study suggests that time-restricted eating may have short-term benefits but long-term adverse effects, and when the study is presented in its entirety it will be interesting and helpful to learn more of the details of the analysis [Note 29].

The three that follow are the details he wants to see. One, nutritional quality — without knowing the nutritional quality of each group's diet, there is no judging whether nutrient density is another explanation. Two, it needs to be emphasised that categorisation into the different time-restricted eating windows was determined on the basis of just two days of dietary intake. Three, he wants a comparison of baseline characteristics across the groups — for example, whether the group with the shortest eating window differed from people following other eating schedules in weight, stress, traditional cardiometabolic risk factors, or other factors associated with adverse cardiovascular outcomes [Note 29].

In one sentence: Gardner is not saying the study is wrong. He is saying that whether the short-window people were already different from everyone else is an unanswered question — residual confounding, in line with the paper's authors' own reservation. And what did not hold up was the all-cause mortality signal, not the cardiovascular one.

⚠ If a headline about fasting raising the risk of cardiac death has made you stop immediately, or if "that study was fake" has made you dismiss it entirely: both reactions run ahead of the evidence. What this study shows is an association, and the authors themselves say they cannot rule out residual confounding — it does not prove that fasting causes cardiovascular death, and it cannot be treated as though it did not exist.

#Who should take particular care — and why this article gives you no "safe list"

⚠ No source has ever published a complete list of people for whom fasting is unsuitable. Every list below is labelled with its source, its date and its scope.

"Not on these lists" does not mean "fasting is safe for you". The four relevant lists cited here differ in how complete they claim to be: three come from Diabetes UK — one of which does not state whether it is exhaustive, another of which says in its own wording that it is giving examples, and a third which hedges every item with "usually" and "for example"; the fourth is a clinical scoring tool from the American Diabetes Association, in which the absence of an item carries no meaning at all. None of the lists below may therefore be treated as exhaustive.

List one: Diabetes UK, "Exemption from fasting". Source: Diabetes and Ramadan | Fasting | Diabetes UK, retrieved 3 August 2026. No review date is displayed on the page itself, but its structured data records a datePublished of 21 August 2017 and a dateModified of 1 July 2026. The full original is at [Note 30]: people can be exempt from fasting if they are children (under the age of puberty), are elderly, are sick or have a certain health condition, have learning difficulties, are travelling, or are pregnant, breastfeeding or menstruating; and anyone showing symptoms of coronavirus, flu or other illness is advised not to fast. ⚠ Scope: this is a list of religious exemptions for Ramadan compiled by a diabetes charity, not a list of medical contraindications for 168 or 5:2 weight-loss fasting. The source does not state whether the list is exhaustive.

List two: Diabetes UK, "Exemptions to fasting" (a different page, explicitly by way of example). Source: Religious fasting and diabetes | Enjoy food | Diabetes UK, structured data recording a dateModified of 24 March 2026. The full original is at [Note 31]: anyone putting their health at serious risk by fasting is usually exempt; for example, people who treat their diabetes with insulin, or who have diabetes complications such as damage to the eyes, kidneys or the nerves in the hands and feet, will not usually be expected to take part in religious fasting; other groups who usually do not have to fast include children, those who are unwell, the elderly and pregnant women. This is the most explicitly non-exhaustive of all the lists cited here — the source itself uses "usually" and "For example" on every item.

List three: Diabetes UK, "Understanding your risk will depend on". From the same Ramadan page, and positioned by the source as what your risk depends on rather than as a list of contraindications. The full original is at [Note 32]: understanding your risk will depend on the type of diabetes you are living with; whether your average blood sugar is currently in a healthy range for you; the type of medication you use to manage your diabetes; whether you take medications that put you at risk of low blood sugar (such as sulphonylureas and insulin); and whether you live with complications such as poor vision, nerve damage, heart or kidney disease (fasting carries a high risk of making these conditions worse).

List four: ADA 2026 table 5.3 — this article deliberately does not reproduce its scores or risk levels. ADA Chapter 5 contains table 5.3 (the components of the prefasting risk calculation for people with diabetes fasting for Ramadan, and the recommended risk scores) and table 5.4 (medication changes during fasting), along with two numbered recommendations, 5.32 and 5.33 [Note 33].

This article will not reproduce the scores, thresholds, risk levels or dose adjustments in those two tables, for the following reason. Table 5.3 is a clinical scoring tool whose components include glomerular filtration rate, glycated haemoglobin values, assessment of complications, and assessments of cognitive function and frailty — all of them information you only have once your medical team has done the tests. It ends in a low-, moderate- or high-risk grading. A reader adding up their own score and concluding they are low risk and fasting is probably safe is a conclusion that cannot be reached at home, and an article should not make it easy to reach. Table 5.4 is a medication dose adjustment table: any change to the timing or dose of insulin or an oral drug can only be decided by your doctor or diabetes team.

Where the line is drawn: what this article will report is how the two tables rank risk — which drug class is flagged as carrying the highest hypoglycaemia risk, which treatment regimen weighs heaviest in the risk calculation (the section on premixed insulin below does exactly this). A ranking of risk is information that sends you to a doctor; a score and a dose is information that tells you not to bother. Only the former is used here. And while the table is printed in full in the ADA original, it is a scoring tool, not a list of contraindications — so a situation not appearing in the table does not make it safe.

What you can do is: know that the tool exists, and at your next appointment ask your medical team directly, "can I have that prefasting risk assessment?"

#Eating disorders: this article reports only the sourced sentence

Popular articles often say that anyone with a history of an eating disorder should not fast. The nearest sourced statement comes from ADA 2026 Chapter 5, and its scope is far narrower than the popular version: caloric restriction may be necessary for glycaemic and weight management, but rigid meal plans and strict tracking of food intake and/or body weight can be contraindicated for individuals at increased risk of significant maladaptive eating behaviours [Note 34].

Note three things: the sentence is written for people with diabetes and prediabetes; it is aimed at rigid meal plans and strict tracking of food or weight; and the wording is "can be contraindicated" — the "can be" is the whole point, not "is contraindicated".

And immediately before it the ADA has a sentence that should be read alongside: along with routine medical management visits, people with diabetes and prediabetes should be screened during diabetes self-management education and support and medical nutrition therapy encounters for a history of dieting and past or current disordered eating behaviours [Note 34]. That sentence runs the opposite way to the first, and is more useful: the first says what people at risk should not do; this one says that this is something clinicians should raise with you.

The Cochrane press release likewise lists underlying eating disorders or behaviours among the factors that make the results non-generalisable [Note 35].

#If you have diabetes, can you fast?

This is the most dangerous practical question in the article, and the three authoritative sources do not say the same thing about it. All three are set out here; this article will not reconcile them for you.

Which is to say: the ADA says that generally it has been shown to be safe for adults with type 1 or type 2 diabetes. But "Generally…has been shown to be safe" — the word "generally" is its own, not an unconditional declaration of safety, and the second half of the same sentence immediately adds that people taking insulin or secretagogues should be medically monitored — but the ADA attaches that monitoring condition only to patients on insulin or secretagogues, not to all adults with diabetes. On the Diabetes UK side, the sentence before Professor Hanif's "can be dangerous" is that it is important to ensure people only fast after discussing it with their diabetes team — that sentence makes no distinction by drug class and applies to everyone with diabetes who wants to fast.

In other words, the two sources differ not only in tone but in scope. And Cochrane 2026 cannot answer the question at all: not one of the included trials reported participants' diabetes status. This article takes the more cautious side: this is not a question to settle on your own.

⚠ The claim that "the highest tier of evidence cannot answer this at all" is not accurate, though, and needs correcting. BMJ 2025 ran its own sensitivity analysis restricted to people with type 2 diabetes (11 studies): compared with ad-libitum, time-restricted feeding showed only a trivial improvement in body weight (−1.93 kg, 95% CI −2.44 to −1.43); whole day fasting (−2.74 kg, −4.30 to −1.18) and continuous energy restriction (−2.87 kg, −4.63 to −1.11) showed small reductions; and alternate-day fasting showed a moderate reduction compared with ad-libitum (−4.42 kg, −5.91 to −2.94) [Note 36].

That analysis is about how many kilograms come off, not about whether it is safe. It has weight figures from 11 type 2 diabetes trials, but it does not answer hypoglycaemia, medication adjustment, or who should not do it.

#⚠ Something that changes how you read the table above: the ADA treats two kinds of fasting separately

ADA Chapter 5 does not have one section on fasting but two: Religious Fasting and Nonreligious Fasting. The ADA sentence quoted in the table above, the one saying it has been shown to be safe, comes from the nonreligious section and is about time-restricted eating; whereas the Diabetes UK page is Diabetes and Ramadan, about the Ramadan fast. In part, they are talking about different things.

That does not make the disagreement disappear, but it changes its nature: it is not two bodies saying opposite things about one thing, but one talking about a religious fast and the other about an eating window. This article will not therefore treat the Diabetes UK sentence as though it did not exist — that page is indeed written for people with diabetes who want to observe the fast.

The ADA sets out the similarities and differences in a figure of its own (Figure 5.1). Three items it lists as common to both: hypoglycaemia risk; prefasting risk assessment and education being essential; and treatment adjustment being required [Note 37]. Four it lists on each side as different: nonreligious fasting has flexible duration and timing, varying levels of intrinsic and extrinsic motivation, hyperglycaemia being less likely as the motive is health related, and no added risk of dehydration; religious fasting has fixed duration and timing, high levels of intrinsic motivation, a risk of hyperglycaemia at the end of fasting hours with or without ketoacidosis, and dehydration being possible in some types [Note 37].

Two things follow, and the reader should hold both.

One, "treatment adjustment is required" is listed by the ADA itself as common to both kinds of fasting. Which is to say you cannot treat your medication as irrelevant on the grounds that you are doing 168 rather than a religious fast. That is not this article's inference; it is what the ADA's own figure says.

Two, but the dehydration risk is not the same. A religious fast means no water either, while time-restricted eating normally allows drinking — the ADA expressly lists "no added risk of dehydration" as a feature of nonreligious fasting. So dehydration warnings attached to religious fasting cannot be carried across unchanged to 168.

⚠ And the medication adjustments in ADA table 5.4 are printed inside the religious fasting section. The table's own title reads "Changes in medications during fasting", with no restriction to one kind of fasting; but the body text that introduces it is about people planning to fast for long hours on multiple consecutive days. So the ADA itself never settles which kind of fasting the table is written for — and that uncertainty is one more reason the matter belongs with your clinical team.

#How the ADA describes the risk

Four named risks: individuals with diabetes who fast have an increased risk of hypoglycaemia, dehydration, hyperglycaemia and ketoacidosis [Note 38].

The ranking of drug classes: for people planning to fast for long hours on multiple consecutive days, the choice of treatment should prioritise drugs with low hypoglycaemia risk; hypoglycaemia risk while fasting in people using insulin, sulfonylureas and other insulin secretagogues is higher than in individuals treated with other types of diabetes medication [Note 38].

A very practical timing point: the timing of glucose monitoring is also especially important, as the last few hours of fasting are frequently associated with approximately 50% of hypoglycaemic events; consequently, avoiding intense physical activity during the last few hours of fasting seems a sensible approach [Note 39].

And the other end of the fast: in one recent study, 16.5% of people with type 2 diabetes who fasted for Ramadan reported high blood glucose above 300 mg/dL (>16.6 mmol/L) on fasting days [Note 39].

The ADA also notes that SGLT2 inhibitors showed no significant change in renal function, dehydration rates or ketosis across several Ramadan studies; the guidance does not recommend changing the SGLT2 inhibitor dose during fasting, but does recommend not starting one close to the beginning of a fast, to avoid excessive thirst.

#How Diabetes UK describes the practicalities

Diabetes UK states: if your blood sugar drops below 4 mmol/l you must break your fast and treat the hypo with sugary food or fluid, following it with something starchy; otherwise you will harm your body and may need medical attention. Carrying hypo treatments and a bottle of water during the fasting period is a good idea [Note 40].

Another sentence is just as important: before choosing to fast, ask your healthcare team what a high blood sugar level is for you; if you go above that level during fasting you must break the fast by drinking water and seek medical advice, since without medical advice this could lead to diabetic ketoacidosis, a serious condition requiring hospital treatment. If you are not able to get hold of your GP or diabetes team, it would be advisable not to fast, especially if you are unsure what to do with your diabetes medications [Note 41].

#⚠ One point where the sources diverge: premixed insulin

The Diabetes UK passage on medication adjustment is a list of what a plan agreed with your healthcare team might include. The lead-in and the item have to be read together: when discussing the risk of fasting with your healthcare team it is important to agree a plan so you can fast safely; the plan may include adjustments to your insulin — you will not need as much insulin before the start of the fast, and the type of insulin may need changing from your usual type; remember, premixed insulin is not recommended during fasting [Note 42].

Note that lead-in: this is not an instruction to the reader to change their own medication, but one item in a plan to be agreed with a healthcare team.

ADA 2026 Chapter 5 puts it differently, though not oppositely: table 5.4 does not tell anyone to avoid premixed insulin, but sets out the direction of timing and dose adjustment for the "Mixed insulin and insulin coformulations" row; and in the "Risk of hypoglycemia" column of that same row, the ADA enters "High" — the same tier as mealtime insulin, the highest in the table. And in the ADA's own prefasting risk calculation tool (table 5.3), "multiple daily injections of premixed insulin" and "once-daily premixed insulin" occupy the two highest risk weightings of all the treatment regimens, above basal insulin, short-acting insulin and every oral drug. (As throughout, this article reproduces none of that table's scores, thresholds or risk levels, only the ranking.)

Which is to say the divergence is narrower than "one says you can and one says you cannot". Diabetes UK says premixed insulin is not recommended during fasting; the ADA does not use the word "avoid", but in both its own tables premixed insulin is flagged at the highest risk tier and calls for a dose reduction. Both point the same way — premixed insulin is the highest-risk class of regimen during fasting; the difference is only that Diabetes UK goes one step further and says "not recommended". This article takes the more cautious side.

Whichever formulation you take, any change to insulin must be decided by your medical team — this article will not and cannot give instructions on it, and reproduces none of table 5.4's adjustments.

#One trial that is often misread

Guo L et al., JAMA Netw Open 2024;7(6):e2416786 (the EARLY trial, nine centres in China, ChiCTR2000040656): 405 adults with newly diagnosed type 2 diabetes randomised to a 5:2 meal replacement group, a metformin group and an empagliflozin group, treated for 16 weeks. At 16 weeks the 5:2 meal replacement group had the largest fall in glycated haemoglobin (least squares mean −1.9%, standard error 0.2%), an adjusted difference of −0.3% against metformin (95% CI −0.4% to −0.1%) and −0.4% against empagliflozin (95% CI −0.6% to −0.2%); on weight, the 5:2 meal replacement group lost 9.7 kg (standard error 2.2), against 5.5 kg for metformin and 5.8 kg for empagliflozin. The authors' own wording is at [Note 43]: 5:2 meal replacement could improve glycaemic outcomes and weight loss in the short term compared with metformin or empagliflozin, making it a promising initial intervention and early management for type 2 diabetes.

Note the authors' word "initial" — a first step, not a step that replaces medication.

⚠ What this trial tested was "5:2 plus a supervised meal replacement programme", in newly diagnosed type 2 diabetes, over 16 weeks, open label, with 332 of the 405 completing (73 lost, 18.0%). It is not a test of someone doing 168 at home, and it is not a reason for anyone to stop their medication. Stopping or changing medication is for your doctor to decide.

#How does this look for people in Hong Kong?

The Centre for Health Protection's obesity and healthy eating pages, and the healthy eating guidelines and information on 「活出健康新方向」, contain nothing at all on intermittent fasting; and the material on fasting on the Hong Kong Dietitians Association's website consists of columns members wrote for newspapers and magazines, not position statements of the association.

#The local figures first, and what they actually cover

The Centre for Health Protection's "Obesity" health topic page (26 April 2023) states that, according to the Department of Health's Population Health Survey 2020-22, 32.6% of people aged 15–84 are obese (body mass index ≥ 25.0 kg/m²) and 22.0% are overweight (23.0 ≤ BMI < 25.0); overweight and obesity are most common among women aged 65–84 (57.0%) and men aged 45–54 (74.6%) [Note 44]. Note: that is "people aged 15–84", not "adults". People aged 15 to 17 are inside that denominator.

Part 2 of the Population Health Survey 2020-22 report itself supplies two further age-standardised figures for people aged 18–84, on two different BMI classifications: on the Asian classification for adults, the age-standardised prevalence of overweight and obesity among people aged 18–84 is 51.3% (crude prevalence 55.1%); on the World Health Organization classification it is 30.8% (crude prevalence 32.8%) [Note 45].

In one sentence: the same people, 51.3% on the Asian classification and 30.8% on the WHO one. Before quoting any figure for how many people in Hong Kong are overweight, you need to know which age range and which set of cut-offs it uses.

And the basis of the figures: the survey covered the land-based non-institutional population of Hong Kong aged 15 or above, excluding foreign domestic helpers and visitors; the physical examinations were conducted between 1 March 2021 and 19 February 2022; and of the respondents who agreed to a physical examination, the numbers selected and invited, and the 2,072 who finally completed the physical examination and blood test (a participation rate of 55.2%), are in the original at [Note 46]. Which is to say: these BMI figures were measured, not self-reported.

#One finding with real local significance

The same report records how people in Hong Kong perceive their own weight: 65.1% of those classified as overweight, 27.6% of those classified as obese class I and 3.2% of those classified as obese class II considered themselves underweight or about right [Note 47]. The same report also states that 54.6% of people aged 15–84 were determined by physical examination to be overweight or obese, comprising 11.2% who self-reported a doctor's diagnosis and 43.3% who did not self-report a doctor's diagnosis but were found on physical examination to have a BMI ≥ 23.0 kg/m² [Note 47].

A contrast: while everyone is debating whether 168 works, 65.1% of those found overweight on physical examination think they are underweight or about right; and 43.3% of people aged 15–84 were found on measurement to be at an overweight or obese BMI but had not self-reported a doctor's diagnosis — a self-report measure, which is not the same as never having been diagnosed. For those people, "which form of fasting" is not the first question.

#What the official pages say, and what they do not

On weight management, the Centre for Health Protection's obesity page has only one sentence:

Maintaining a balanced diet and taking regular physical activity are effective in achieving and maintaining a suitable body weight. (our translation from the Chinese original)

Chinese original:

「保持均衡飲食和定期進行體能活動能有效達至及保持適中的體重。」

As for how much physical activity, note where that recommendation comes from. The Centre for Health Protection's "Physical Activity" topic page (15 April 2025) states that the World Health Organization recommends that adults aged 18 and over do at least 150–300 minutes of moderate-intensity aerobic physical activity a week, or at least 75–150 minutes of vigorous-intensity aerobic activity, or an equivalent combination; and that they should also do muscle-strengthening activities targeting all major muscle groups on two days or more each week [Note 48]. These figures are the Centre for Health Protection quoting the WHO's recommendation, not the Department of Health's own. The same page also records that the Behavioural Risk Factor Survey of 2023 found 14.8% of people aged 18 and above to be insufficiently physically active (16.0% of women, 13.4% of men) — note that this is "aged 18 and above", a different basis from the "15–84" of the obesity figures above.

#What is and is not on the official pages

Each item below states which page was checked and which version, so that you can open it yourself.

Centre for Health Protection "Obesity" topic page (Chinese and English versions, page-dated 26 April 2023): the whole page has one sentence on weight management and then directs the reader to the 「活出健康新方向」 website. Neither language version contains any of the words 斷食, 間歇, 限時, "fasting" or "intermittent".

Centre for Health Protection "Healthy Eating" resources page: the page lists 35 links, and setting aside 3 duplicate "accessible version" entries, 32 publications, including 「你是否過胖」, 「中央肥胖」, 「肥胖危機」, 「熱量知多少」 and five healthy eating pyramids by age group. None of the 32 has fasting as its subject; the two booklets 「你是否過胖」 and 「熱量知多少」 and the 「中央肥胖」 poster were each opened individually, and their text does not mention fasting either.

Change4Health, "Healthy Eating" section: under "Guidelines" there are only four items — nutritional guidelines for children in Hong Kong, the healthy eating food pyramid for Hong Kong, joint recommendations on iron intake for the public (women of childbearing age above all), and joint recommendations on iodine intake; under "Facts" there are four; under FAQs there are six. None of them concerns fasting.

One signpost worth following: under "Healthy Eating" in Change4Health's "Related Links (Local)", the official pages link out to the Hong Kong Dietitians Association and the Hong Kong Nutrition Association — which is to say that the Government itself directs readers to these two professional bodies.

#There is fasting material on the Hong Kong Dietitians Association's site — but not the kind you think

The "Dietitians on Media" section of the Hong Kong Dietitians Association's website is indexed under twelve subject headings holding 533 records in total, of which the single heading on obesity and the three highs holds 33. Three of them are about intermittent fasting.

Three things have to be said together, because they settle how much weight this material carries.

One, this index is a record of what members have written in the media, not a set of documents published by the association. Every entry clicks through to the original URL at a newspaper or magazine, not to a page of the association's own.

Two, neither the association's website nor its sitemap has a section for position statements or guidelines. The sitemap lists "About HKDA", "About Dietitians", "Membership", "Find a Dietitian", "Resources", "Dietitians on Media" and "Contacts" — not one of them is a professional position document.

Three, the most recent of the three is from 2022, and the original publications of the other two have both closed. April 2022 is before every one of BMJ 2025, Cochrane 2026 and ADA 2026 appeared.

⚠ If you search for Hong Kong dietitians and intermittent fasting and take the association's name as an official position: what the index holds is members' personal newspaper columns, not the association's position; and the few entries you will find were written before all three of the main documents cited here existed. (Incidentally, the 2022 title uses the characters 「間竭性斷食」 — the second character is not 「歇」 — so searching the association's site for 「間歇」 will not find it.)

#One limit on generalising, and one exception

The Cochrane press release notes that the majority of the included studies enrolled predominantly white populations in high-income countries [Note 49].

The exception is good news: both of the key trials cited here were conducted in China — Liu D et al.'s 12-month trial of 2022 at Nanfang Hospital in Guangzhou, and the EARLY trial across nine centres nationwide — so the finding that time-restricted eating plus calorie restriction is not superior to calorie restriction alone at one year was obtained in a Chinese population.

#What to do next

1. Ask what the control group was for whatever figure you are reading. Against eating as usual, against regular dietary advice, against the same calorie deficit — three completely different answers, and all three are true.
2. Ask which form of fasting was tested, and for how long. The evidence for the three forms is not interchangeable; 24 weeks is the watershed, and there are only five trials of 52 weeks or longer.
3. Do not fast for your lipids or your blood glucose. At the same calorie deficit, no benefit was observed for glycated haemoglobin or HDL cholesterol in any comparison, and LDL moved upwards in the one-year alternate-day fasting trial.
4. If you have diabetes, talk to your diabetes team before deciding. The specific things to ask: which of my current medications carry a high hypoglycaemia risk (insulin, sulfonylureas, other insulin secretagogues)? How often should I monitor my blood glucose while fasting? At what blood glucose must I break the fast immediately?
5. Ask your medical team whether they can do a prefasting risk assessment. That tool needs eGFR, glycated haemoglobin, an assessment of complications and an assessment of frailty — it is something to request at an appointment, not to score for yourself at home.
6. Do not adjust your own insulin. Both sources flag premixed insulin as the highest-risk class during fasting, and any change of dose or timing can only be decided by your medical team.
7. If your goal is weight loss, work out which box you are in first. Hong Kong's measured data show that 65.1% of those classified as overweight consider themselves about right — for those people, "which form of fasting" is not the first question.
8. Do not use the "who should not fast" lists circulating online to rule yourself out. None of the four lists cited here claims to be exhaustive, and three of them are religious exemption lists, not lists of medical contraindications.
9. If you see a headline about fasting raising the risk of cardiac death, hold two things in mind: the paper has now been formally published (so the "not peer reviewed" retort no longer works), but it is an observational study, and its authors say themselves that they cannot rule out residual confounding.

#Frequently asked questions

So does 168 work for me or not?

It depends what you are comparing it with. Against "change nothing and carry on eating as before", starting a form of intermittent fasting produces weight loss on average (the Nutrition Journal meta-analysis of 2025 reports −3.73 kg, 95% CI −5.29 to −2.17, against a control group maintaining habitual dietary patterns). Against "the same calorie deficit", the BMJ 2025 network meta-analysis finds no difference in favour of intermittent fasting over continuous energy restriction in trials of 24 weeks or more — but the authors state themselves that this "not seen" may reflect an insufficient number of long enough studies rather than a demonstrated absence. This article offers no view on whether you personally should fast.

Is alternate-day fasting the strongest form?

In BMJ 2025, alternate-day fasting is the only strategy showing a weight difference against continuous energy restriction (−1.29 kg, 95% CI −1.99 to −0.59). But the same authors note that this figure falls below their own minimally important difference of 2.0 kg and describe the clinical impact in those comparisons as limited; and in Trepanowski's one-year trial of 2017, the alternate-day fasting group had the highest dropout of the three (13 of 34, 38%) and LDL cholesterol at month 12 was 11.5 mg/dL higher than in the daily calorie restriction group.

Is finishing earlier in the day — before three in the afternoon, say — better?

The strongest single positive trial at present is Jamshed H et al. (JAMA Intern Med 2022;182(9):953-962): 90 people over 14 weeks, both groups on the same energy restriction, with the fasting group's window running from 7am to 3pm. Results: 2.3 kg more weight lost (95% CI −3.7 to −0.9, P = .002), no difference in body fat in the analysis of all randomised participants (−1.4 kg, 95% CI −2.9 to 0.2, P = .09), and diastolic blood pressure 4 mm Hg lower (95% CI −8 to 0, P = .04); the authors describe the effect as equivalent to a further 214 kcal a day. ⚠ But the next part of the same abstract has to be read with it: a prespecified completers analysis (59 people) instead found 1.8 kg more body fat lost (95% CI −3.6 to 0.0, P = .047) and 1.2 kg more trunk fat (95% CI −2.2 to −0.1, P = .03). The two analyses point in opposite directions: the analysis of all randomised participants is the more robust one, a completers analysis is biased towards those who kept it up, and its confidence interval touches zero at the upper end. Both were printed by the authors, and both are reported here. Note also that what was tested is 07:00–15:00, against a control group with an eating window of 12 hours or more a day — not the 12:00–20:00 style of 168 that people in Hong Kong actually do.

Can I just restrict the timing without counting calories?

That is exactly what the TREAT trial tested (Lowe DA et al., JAMA Intern Med 2020;180(11):1491-1499): 116 people over 12 weeks, a 16:8 eating window running from 12 noon to 8pm, with no calorie prescription in either group. The between-group weight difference was −0.26 kg (95% CI −1.30 to 0.78, P = .63), with no difference in estimated energy intake either. The authors' conclusion: "Time-restricted eating, in the absence of other interventions, is not more effective in weight loss than eating throughout the day."

Does fasting cost you muscle?

There is a small but consistently directed signal. Schroor 2024 pooled 28 trials (2 to 52 weeks) and found intermittent energy restriction reduced fat-free mass by 0.20 kg more than continuous energy restriction (95% CI −0.39 to −0.01); the in-person visit cohort of the TREAT trial (50 people) likewise found a between-group difference in appendicular lean mass index of −0.16 kg/m² (95% CI −0.27 to −0.05). Schroor's authors add themselves that energy intake may not have been equal between study groups, which may have affected the result.

I have diabetes — can I just try it?

This article will not answer that, because the three authoritative sources do not agree on the point (see the table above), and not one of the 22 trials included in Cochrane 2026 reported participants' diabetes status. What can be said is: the ADA notes that people using insulin, sulfonylureas and other insulin secretagogues face a higher hypoglycaemia risk while fasting and should be medically monitored during the fasting period; and Diabetes UK says that if you cannot get hold of your GP or diabetes team, it would be advisable not to fast. Any medication adjustment can only be decided by your medical team.

#Notes: the original texts

[Note 1] ADA Standards of Care in Diabetes—2026, Chapter 5: "Intermittent fasting is an umbrella term that includes three main forms of restricted eating: alternate-day fasting (energy restriction of 500–600 calories on alternate days), the 5:2 eating pattern (energy restriction of 500–600 calories on consecutive or nonconsecutive days with usual intake the other five), and time-restricted eating (daily calorie restriction based on window of time of 8–15 h)."

[Note 2] Cochrane 2026 inclusion criteria: "We included randomised controlled trials (RCTs) and cluster-RCTs that compared intermittent fasting (including time-restricted feeding, periodic fasting, alternate-day fasting, and modified alternate-day fasting) with regular dietary advice, no intervention or waiting list in men and women with overweight or obesity, with or without associated comorbid conditions. The minimum duration of the intervention was four weeks, and the minimum duration of follow-up was six months. We excluded cross-over and quasi-RCTs."

[Note 3] ADA 2026 Chapter 5, "Nonreligious Fasting": "Each produces mild to moderate weight loss (3–8% loss from baseline) over short durations (8–12 weeks) with no significant differences in weight loss when compared with continuous calorie restriction."

[Note 4] The same, citing a pooled analysis of 2024: "The authors concluded that intermittent energy restriction produces small but significant reductions in waist circumference and fat-free mass but were otherwise not superior to continuous energy restriction eating patterns."

[Note 5] Cochrane 2026: "Compared to regular dietary advice, intermittent fasting may result in little to no difference in percentage from baseline weight loss (MD −0.33, 95% CI −0.92 to 0.26; 21 studies, 1430 participants; low-certainty evidence due to risk of bias)."

[Note 6] BMJ 2025: "All intermittent fasting and continuous energy restriction diet strategies reduced body weight when compared with ad-libitum diet. Compared with continuous energy restriction, alternate day fasting was the only form of intermittent fasting diet strategy to show benefit in body weight reduction (mean difference −1.29 kg (95% CI −1.99 to −0.59), moderate certainty of evidence)."

[Note 7] Parkinson M, Cochrane news, 16 February 2026: "Intermittent fasting is unlikely to lead to greater weight loss in overweight or obese adults than traditional dietary advice or doing nothing, a new Cochrane review finds."

[Note 8] BMJ 2025: "Estimates were similar among trials with less than 24 weeks follow-up (n=76); however, moderate-to-long-term trials (≥24 weeks, n=17) only showed benefits in weight reduction in diet strategies compared with ad-libitum." And "In trials of 24 weeks or more (n=17), the most restrictive strategies (ADF, TRE, and CER) showed small body weight reductions versus ad-libitum (mean difference range −1.88 to −3.63 kg), with no differences between intermittent fasting strategies and CER in these moderate-to-long term studies."

[Note 9] The same, discussion: "The loss of association in the network assessment of moderate to longer term trials (≥24 weeks) may be due to an insufficient number of studies available."

[Note 10] The same, legend to figure 4: "The minimally important difference for body weight was 2.0 kg." Discussion: "While many observed reductions with intermittent fasting strategies compared to ad-libitum exceeded this threshold, individual strategies of intermittent fasting particularly ADF versus CER (−1.29 kg) fell below this level, suggesting somewhat limited clinical impact in those comparisons."

[Note 11] The same: "Notably, the results show that intermittent fasting may offer unique benefits primarily in the short term, whereas both intermittent fasting and CER appear to provide similar moderate-to-long term improvements over ad-libitum diets. This equivalence in sustained outcomes in the moderate-to-long term is a critical take-away for clinicians managing chronic metabolic conditions."

[Note 12] Liu D et al., N Engl J Med 2022;386(16):1495-1504: "Among patients with obesity, a regimen of time-restricted eating was not more beneficial with regard to reduction in body weight, body fat, or metabolic risk factors than daily calorie restriction."

[Note 13] Trepanowski JF et al., JAMA Intern Med 2017;177(7):930-938: "Alternate-day fasting did not produce superior adherence, weight loss, weight maintenance, or cardioprotection vs daily calorie restriction."

[Note 14] The same: "Among the 100 participants (86 women and 14 men; mean [SD] age, 44 [11] years), the dropout rate was highest in the alternate-day fasting group (13 of 34 [38%]), vs the daily calorie restriction group (10 of 35 [29%]) and control group (8 of 31 [26%])." And "Participants in the alternate-day fasting group ate more than prescribed on fast days, and less than prescribed on feast days, while those in the daily calorie restriction group generally met their prescribed energy goals."

[Note 15] The same, limitations: "Third, since the dropout rate was higher than anticipated, our power to detect the hypothesized difference of 5% weight loss between the intervention groups at month 6 decreased from 80% to 60%. The higher dropout rate in the alternate-day fasting group may have also introduced a possible selection bias between groups."

[Note 16] BMJ 2025: "Of the 99 identified studies, 74 studies included description of adherence or sufficient information to determine adherence. Most trials reported high adherence (>80%) throughout the duration of the study. Among the trials that had a 52 week follow-up, most stated poor adherence in the intervention groups. In one study, adherence to the WDF declined from 74% at six weeks to 22% at 52 weeks. By contrast, however, another 52 week study reported high adherence in the TRE and CRE diet groups (both approximately 84%)."

[Note 17] BMJ 2025: "Additionally, the scarcity of trials of 52 weeks or longer (only five identified) precluded a separate network meta-analysis for very long term effects, limiting relevance to related to sustained long term weight loss." Cochrane 2026: "The included studies focused on short-term effects of the intervention (up to 12 months), limiting the applicability of the evidence in this review to inform decision-making for longer durations."

[Note 18] BMJ 2025: "No differences were noted between intermittent fasting, continuous energy restriction, and ad-libitum diets for HbA1c and high density lipoprotein." "ADF was the only intermittent fasting strategy to show an improvement in anthropometric and lipid measures when compared to CER. No benefit was observed for HbA1c or high density lipoprotein cholesterol in any diet strategy comparison." And "time restricted eating resulted in a small increase in total cholesterol, low density lipoprotein cholesterol, and non-high density lipoprotein cholesterol"

[Note 19] Trepanowski 2017: "Mean low-density lipoprotein cholesterol levels were significantly elevated by month 12 among the participants in the alternate-day fasting group (11.5 mg/dL [95% CI, 1.9-21.1 mg/dL]) compared with those in the daily calorie restriction group."

[Note 20] The same, discussion: "However, the trial included primarily metabolically healthy obese adults. Since many of the participants had normal cholesterol levels and normal blood pressure at baseline, it is not surprising that most risk indicators for cardiovascular disease did not change in response to diet."

[Note 21] Schroor MM et al., Adv Nutr 2024;15(1):100130: "All IER diets combined reduced fat-free mass (WMD: –0.20 kg; 95% CI: –0.39 to –0.01; P = 0.044) and waist circumference (WMD: –0.91 cm; 95% CI: –1.76 to –0.06; P = 0.036) more than CER. Effects on body mass index [BMI (kg/m2)], glucose, insulin, homeostatic model assessment for insulin resistance (HOMA-IR), serum lipid and lipoprotein concentrations, and blood pressure did not differ. Further, TRE reduced body weight, FM, and fat-free mass more than CER, whereas ADF improved HOMA-IR more. BMI was reduced less in the 5:2 diet compared with CER." The addition immediately following the conclusion sentence: "Slightly greater reductions were, however, observed in fat-free mass and waist circumference."

[Note 22] BMJ 2025: "Most trials reported mild side effects including constipation, nausea, hunger, diarrhoea, and dizziness."

[Note 23] Cochrane press release: "Reporting of side effects was inconsistent across trials, making it difficult to draw firm conclusions. The evidence base remains limited, with only 22 trials, many with small sample sizes and inconsistent reporting."

[Note 24] Titmus M, signed review on the British Dietetic Association website, 28 September 2021: "No serious adverse implications of TRE were reported in the literature. Mild events such as headaches, diarrhoea, increased thirst and dizziness were potentially related to TRE intervention, but occurred in the early stages and resolved by week three following adjustment to TRE."

[Note 25] The same: "TRE may inadvertently exacerbate existing vulnerability to overeating." And immediately after: "Of note is that participants ultimately realised that they did not experience excessive hunger and discontinued deliberate overeating behaviour." The same section also carries: "Nevertheless, individuals considering TRE will need to be informed of potential transient effects to support adherence."

[Note 26] American Heart Association newsroom, 18 March 2024: "People who followed a pattern of eating all of their food across less than 8 hours per day had a 91% higher risk of death due to cardiovascular disease." The same release: "Abstracts presented at the Association's scientific meetings are not peer-reviewed, rather, they are curated by independent review panels… The findings are considered preliminary until published as a full manuscript in a peer-reviewed scientific journal." The senior author: "Although the study identified an association between an 8-hour eating window and cardiovascular death, this does not mean that time-restricted eating caused cardiovascular death."

[Note 27] Chen M, Xu L, Van Horn L et al., Diabetes Metab Syndr 2025;19(7):103278: "Adult participants who reported usual intake from two valid 24-h dietary recalls were included from the National Health and Nutrition Examination Survey in 2003-2018 (n = 19,831)… During a median follow-up of 8.1 years, compared with eating duration of 12-14 h, eating duration <8 h was robustly associated with higher cardiovascular mortality (HR, 2.35 [95 % CI, 1.39-3.98]), but not with all-cause and cancer mortality. The positive association with cardiovascular mortality remained consistent across 8 subgroups stratified by race/ethnicity, socioeconomic factors, and smoking status, and survived 14 sensitivity analyses. However, the association with all-cause mortality did not survive many sensitivity analyses."

[Note 28] The same, conclusion: "Although a positive association was observed between eating duration <8 h and cardiovascular mortality, further research is required to understand whether this risk is attributed to the short eating duration itself or residual confounding resulting from its contributing factors."

[Note 29] Comment by Dr Christopher D. Gardner: "Overall, this study suggests that time-restricted eating may have short-term benefits but long-term adverse effects. When the study is presented in its entirety, it will be interesting and helpful to learn more of the details of the analysis." "Second, it needs to be emphasized that categorization into the different windows of time-restricted eating was determined on the basis of just two days of dietary intake," "It will also be critical to see a comparison of demographics and baseline characteristics across the groups that were classified into the different time-restricted eating windows – for example, was the group with the shortest time-restricted eating window unique compared to people who followed other eating schedules, in terms of weight, stress, traditional cardiometabolic risk factors or other factors associated with adverse cardiovascular outcomes?"

[Note 30] Diabetes UK, Diabetes and Ramadan: "Exemption from fasting — People can be exempt from fasting if they: are children (under the age of puberty); are elderly; are sick or have a certain health condition; have learning difficulties; are travelling; are pregnant, breastfeeding or menstruating. If you're showing any symptoms of coronavirus (Covid-19), flu or other illness, it would be advisable not to fast. If you're fasting during Ramadan, getting the coronavirus vaccine does not break your fast."

[Note 31] Diabetes UK, Religious fasting and diabetes: "Exemptions to fasting — Anyone who is putting their health at serious risk by fasting is usually exempt (doesn't have to take part). For example, you won't usually be expected take part in religious fasting if you treat your diabetes with insulin or have diabetes complications such as damage to your eyes, kidneys or nerves in your hands and feet. Speak to your healthcare team for advice if you aren't sure if you should fast and your spiritual leader for guidance. Other groups that usually don't have to fast include children, those who are unwell, the elderly and pregnant women." The same page also carries: "If you are unwell or have any symptoms of illness such as a cold, Covid-19 or flu, it's best not to fast."

[Note 32] Diabetes UK, Diabetes and Ramadan: "Risks of fasting — It is important to discuss with your healthcare team how living with diabetes and following Ramadan could put your health at risk. Understanding your risk will depend on: The type of diabetes you are living with. If you are currently keeping your average blood sugar level (HbA1c) in a healthy range for you. The type of medication you use to manage your diabetes. If you take medications that put you at risk of low blood sugar (also called hypoglycaemia), such as sulphonylureas and insulin. If you're living with diabetes complications such as poor vision, nerve damage, heart or kidney disease. There is a high risk that fasting could make these health conditions worse. For example, people living with type 2 diabetes who manage with diet and lifestyle only, or who take one diabetes medication that does not increase their risk of hypoglycaemia will have a lower risk during the fast if they are already keeping their HbA1c in a healthy range."

[Note 33] ADA 2026 Chapter 5, recommendations 5.32 and 5.33: "5.32 Use the updated International Diabetes Federation along with Diabetes and Ramadan International Alliance comprehensive prefasting risk assessment to generate a risk score for the safety of religious fasting. Provide fasting-focused education to minimize risks. B" "5.33 Assess and optimize treatment plan, dose, and timing for people with diabetes well in advance of religious fasting to reduce risk of hypoglycemia, dehydration, hyperglycemia, and/or ketoacidosis. B"

[Note 34] ADA 2026 Chapter 5: "Caloric restriction may be necessary for glycemic and weight management, but rigid meal plans and strict tracking of food intake and/or body weight can be contraindicated for individuals at increased risk of significant maladaptive eating behaviors." Immediately before it: "Along with routine medical management visits, people with diabetes and prediabetes should be screened during DSMES and MNT encounters for a history of dieting and past or current disordered eating behaviors."

[Note 35] Cochrane press release: "The authors therefore warn that these results may provide clues, but cannot be extrapolated to the entire population, as they may vary depending on sex, age, ethnic origin, disease status, or underlying eating disorders or behaviours."

[Note 36] BMJ 2025 type 2 diabetes sensitivity analysis: "We conducted a sensitivity network meta-analyses for studies with people with type 2 diabetes (n=11)… TRF compared with ad-libitum only showed a trivial improvement for body weight (mean difference −1.93 kg (95% CI −2.44 to −1.43)), while WDF (−2.74 kg (−4.30 to −1.18)) and CER (−2.87 kg (−4.63 to −1.11)) diets showed small reductions in body weight compared with ad-libitum. ADF, however, showed a moderate reduction in body weight compared with ad-libitum (−4.42 kg (−5.91 to −2.94))."

[Note 37] ADA 2026 Figure 5.1, the three items common to both: "Hypoglycemia risk / Risk assessment and education is essential prefasting / Treatment adjustment is required"; the four for nonreligious fasting: "Flexible duration and timing" "Varying levels of intrinsic and extrinsic motivation" "Hyperglycemia is less likely as motive is health related" "No added risk of dehydration"; the four for religious fasting: "Fixed duration and timing" "High levels of intrinsic motivation" "Risk of hyperglycemia at end of fasting hours with or without ketoacidosis" "Dehydration is possible in some types".

[Note 38] ADA 2026 Chapter 5: "Individuals with diabetes who fast have an increased risk for hypoglycemia, dehydration, hyperglycemia, and ketoacidosis." And "for people planning to fast for long hours and multiple consecutive days, choice of treatment should prioritize drugs with low hypoglycemia risk. Hypoglycemia risk while fasting in people using insulin, sulfonylureas, and other insulin secretagogues is higher than that for individuals treated with other types of diabetes medications."

[Note 39] The same: "Timing of glucose monitoring is also especially important, as the last few hours of fasting are frequently associated with approximately 50% of hypoglycemic events. Consequently, avoiding intense physical activity during the last few hours of fasting seems to be a sensible approach." And "in one recent study 16.5% of people with type 2 diabetes who fasted for Ramadan reported high blood glucose of >300 mg/dL (>16.6 mmol/L) during fasting days."

[Note 40] Diabetes UK: "If your blood sugars drop below 4 mmol/l you must break your fast and treat the hypo with some sugary food or fluid and follow this with something starchy as otherwise you will harm your body and may need medical attention. It's a good idea to carry hypo treatments with you and a bottle of water during the fasting period."

[Note 41] The same: "Before choosing to fast, ask your healthcare team what a high blood sugar level is for you. If you go above that level during fasting, you must break the fast by drinking water and seek medical advice. Without medical advice, this could lead to diabetic ketoacidosis (DKA) – a serious condition requiring hospital treatment. If you're not able to get hold of your GP or diabetes team, it would be advisable not to fast, especially if you're not sure of what to do with your diabetes medications."

[Note 42] The same: "When discussing the risk of fasting with your healthcare team it is important to agree a plan so you can fast safely. This may include: … Adjustments to your insulin. You won't need as much insulin before the start of the fast. Also, the type of insulin may need changing from your usual type. Remember, pre-mixed insulin is not recommended during fasting."

[Note 43] Guo L et al., JAMA Netw Open 2024;7(6):e2416786: "found that 5:2 MR could improve glycemic outcomes and weight loss in the short term compared with metformin or empagliflozin, making it a promising initial intervention and early management for type 2 diabetes."

[Note 44] Centre for Health Protection "Obesity" health topic page (26 April 2023):

According to the Population Health Survey 2020-22 conducted by the Department of Health, among people aged 15–84, 32.6% (26.4% of women and 39.4% of men) are obese (that is, a body mass index ≥ 25.0 kg/m2), and 22.0% (19.7% of women and 24.6% of men) are overweight (that is, 23.0 kg/m2 ≤ body mass index < 25.0 kg/m2). Overweight and obesity are most common among women aged 65–84 (57.0%) and men aged 45–54 (74.6%). (our translation from the Chinese original)

Chinese original:

「據衞生署進行的2020-22年度人口健康調查顯示，在15－84歲人士中，有32.6%的人士(女性為26.4%，男性為39.4%)屬於肥胖(意即體重指標 ≥ 25.0公斤／米2)，有22.0%的人士(女性為19.7%，男性為24.6%)為超重(意即23.0公斤／米2 ≤ 體重指標 < 25.0公斤／米2)。超重及肥胖最常見於65－84歲的女性(57.0%)和45－54歲的男性(74.6%)。」

[Note 45] Part 2 of the Population Health Survey 2020-22 report:

On the weight classification for Asian adults (that is, overweight at a body mass index ≥ 23.0 kg/m2 and obesity at ≥ 25.0 kg/m2), the age-standardised prevalence of overweight and obesity among people aged 18–84 was 51.3% (crude prevalence 55.1%). On the World Health Organization's weight classification (that is, overweight at a body mass index ≥ 25.0 kg/m2 and obesity at ≥ 30.0 kg/m2), the age-standardised prevalence among people aged 18–84 was 30.8% (crude prevalence 32.8%). (our translation from the Chinese original)

Chinese original:

「根據亞洲成年人的體重分類（即超重體重指標 ≥ 23.0 kg/m2；肥胖體重指標 ≥ 25.0 kg/m2）， 18－84 歲人士的年齡標準化超重及肥胖普遍率為51.3%（粗略普遍率為55.1%）。而根據世界衞生組織的體重分類（即超重體重指標 ≥ 25.0 kg/m2；肥胖體重指標 ≥ 30.0 kg/m2），18－84 歲人士的年齡標準化超重及肥胖普遍率為30.8%（粗略普遍率為32.8%）。」

[Note 46] The same:

This survey covered the land-based non-institutional population of Hong Kong aged 15 or above, excluding foreign domestic helpers and visitors. The physical examinations were conducted between 1 March 2021 and 19 February 2022… Of the 6 373 respondents who agreed to a physical examination, 3 757 were selected and invited to book a health check, of whom 2 072 completed the physical examination and blood test (a participation rate of 55.2%) (our translation from the Chinese original)

Chinese original:

「本調查涵蓋香港年滿15 歲或以上的陸上非住院人口，不包括外籍家庭傭工和訪港旅客。身體檢查則於2021 年3 月1 日至2022 年2 月19 日期間進行……在同意接受身體檢查的6 373 名受訪者中，有3 757 名受訪者獲抽選及邀請預約進行健康檢查，當中有2 072 名受訪者完成體格檢查和血液檢驗（參與率為55.2%）」

[Note 47] The same:

65.1% of those classified as overweight, 27.6% of those classified as obese class I and 3.2% of those classified as obese class II considered themselves underweight or of suitable weight ("about right"). (our translation from the Chinese original)

Chinese original:

「有 65.1%被歸類為超重的人士、27.6%被歸類為一級肥胖的人士及 3.2%被歸類為二級肥胖的人士認為他們自己過輕或體重適中（「剛好」）。」

And:

54.6% of people aged 15–84 (46.1% of women and 64.0% of men) were determined by physical examination to be overweight or obese, comprising 11.2% (10.6% of women and 11.9% of men) who self-reported a doctor's diagnosis of overweight or obesity, and 43.3% (35.4% of women and 52.1% of men) who did not self-report a doctor's diagnosis of overweight or obesity but were found at physical examination to have a body mass index ≥ 23.0 kg/m2, that is, overweight or obese. (our translation from the Chinese original)

Chinese original:

「有54.6%的15－84歲人士（女性為46.1%，男性為64.0%）透過體格檢查定為超重或肥胖，包括11.2%（女性為 10.6%，男性為 11.9%）有自述經醫生診斷超重或肥胖，及 43.3%的人士（女性為 35.4%，男性為 52.1%）沒有自述經醫生診斷超重或肥胖，但在體格檢查時發現體重指標 ≥ 23.0 kg/m2，即超重或肥胖。」

[Note 48] Centre for Health Protection "Physical Activity" health topic page (15 April 2025):

For adults aged 18 and above (including those living with a chronic condition or disability), the World Health Organization recommends that they should undertake regular physical activity. They should do at least 150–300 minutes of moderate-intensity aerobic physical activity a week, or at least 75–150 minutes of vigorous-intensity aerobic physical activity, or at least an equivalent amount of a combination of moderate- and vigorous-intensity activity, for substantial health benefits. They should also do muscle-strengthening activities at moderate or greater intensity targeting all major muscle groups on two days or more each week, which brings additional health benefits. (our translation from the Chinese original)

Chinese original:

「對18歲及以上（包括患有慢性病或殘疾）的成年人來說，世界衞生組織建議：他們應定期進行體能活動。他們應每星期進行最少150–300分鐘中等強度的帶氧體能活動；或最少75–150分鐘劇烈強度的帶氧體能活動；或最少相等於混合中等和劇烈強度活動模式的時間，以獲得顯著健康裨益。他們還應每星期有兩天或以上，進行中等或更高強度針對所有主要肌肉群的強化肌肉活動，這能帶來額外健康裨益。」

The same page also records that the single sentence on weight management on the Centre for Health Protection's obesity page is 「保持均衡飲食和定期進行體能活動能有效達至及保持適中的體重。」

[Note 49] Cochrane press release: "The majority of the included studies enrolled predominantly white populations in high-income countries."

#What this article does not state

Each of the following is something this article could have written but has chosen not to, for want of sufficient sourcing:

1. "Anyone with a history of an eating disorder should not fast" — this article does not state it in that unbounded form. The sourced statement on the point comes from ADA 2026 and consists of two sentences, both reported here: one restricted to people with diabetes and prediabetes, restricted to rigid meal plans and strict tracking of food or weight, and using "can be contraindicated" rather than "is contraindicated"; and the sentence immediately before it, requiring clinicians to screen for a history of dieting and disordered eating behaviours during diabetes self-management education and medical nutrition therapy encounters. Neither says that a history of an eating disorder means one should not fast.
2. "You should not fast while breastfeeding" — not written. This item appears only within Diabetes UK's religious exemption list, and no clinical source supports treating it as a medical contraindication.
3. "You should not fast while pregnant" — not written in generalised form. The strongest sourced statement is that the ADA 2026 risk assessment tool (table 5.3) lists pregnancy with diabetes as one of its risk elements, and that tool is designed for a clinician to score against a patient's overall situation before judging a risk level. This article reproduces none of that table's scores, thresholds or risk levels — including this item — for the reason given above: the tool needs eGFR, glycated haemoglobin, assessment of complications and assessment of frailty, none of which this article's readers have, and working out a score for yourself is precisely what this article is avoiding. A pregnant woman with diabetes considering fasting should have that assessment done with her clinical team. But its scope is pregnant women with diabetes, fasting for Ramadan — not pregnancy in general, and not 168 for weight loss.
4. "Older people should not fast because of the risk of sarcopenia" — not written. The only source that mentions sarcopenia (Titmus 2021) lists it as a direction for further research, not a contraindication.
5. "Some medicines have to be taken with food, so fasting is impossible" — not written. This article found no source dealing with the relationship between non-diabetes medications and the timing of eating.
6. "Adolescents should not fast" — not written as a clinical contraindication. "Children" appears only within two religious exemption lists. ISPAD published a consensus guideline in 2022 on fasting for young people with diabetes (Pediatr Diabetes 2022;23(8):1512-1528), but this article could not obtain its content and will therefore not describe what it says.
7. The LDL figure from the Nutrition Journal 2025 analysis — not cited here. That abstract prints an identical 95% confidence interval (−12.36, −0.26) for both total cholesterol (−6.31 mg/dl) and LDL (−5.44 mg/dl); at least one of the two is a misprint, which is reported here as found rather than corrected, and the LDL value is not used.
8. The phrase "all but one was considered severe" from the BMJ 2025 adverse events passage — not cited here. That sentence contradicts the two immediately following it (describing most side effects as mild and naming a single serious event). This article uses only the specific, non-contradictory facts in that passage.
9. The claim circulating that "about 65% of the weight lost on time-restricted eating is lean mass" — not used here. ⚠ Correction: that figure does not come from the Titmus review of 2021 but from the TREAT trial paper itself (Lowe 2020, discussion: the in-person visit cohort's time-restricted eating group lost a mean of 1.70 kg, of which 1.10 kg — about 65% — was lean mass, against a normal 20% to 30% of lean mass in weight loss). The reason it is not used is that the TREAT authors state in their own limitations that protein intake was not measured, and that dual-energy X-ray absorptiometry does not account for muscle hydration status; and in restating it Titmus 2021 adds that physical activity fell in the fasting group and writes "Whether TRE directly influences muscle loss is questionable". This article uses Schroor 2024's pooled figure with a confidence interval instead.
10. Whether the Hospital Authority or local Chinese medicine bodies have any relevant material — this article states nothing either way (neither that they do nor that they do not).
11. ⚠ Correction: the Cochrane 2026 review does have a traditional Chinese version, and this article's earlier statement that this was unknown was wrong. It is at https://www.cochrane.org/zh-hant/evidence/CD015610_intermittent-fasting-adults-overweight-or-obesity, translated by Lin Yun-ting and coordinated by the Cochrane Taiwan centre at Taipei Medical University and the East Asian Cochrane Alliance. But this article still does not put Chinese renderings of Cochrane sentences inside quotation marks, for reasons that have changed: first, that version translates the plain-language summary, not the scientific abstract this article cites; and second, the page itself states that where the translation differs from the original, is inaccurate or is inappropriate, the English original prevails. A translation that declares the English authoritative cannot be cited as the original. (That Chinese version also reproduces the internal inconsistency identified above under "Two gaps the reader is owed": its plain-language summary says there may be little to no difference for weight loss, while its limitations paragraph says a certain effect was achieved and that they have some confidence in the result.)
12. ⚠ Correction: the two errata to the TREAT trial have now been obtained, and neither affects any figure cited here. The first (JAMA Intern Med 2020;180(11):1555) corrects the data alignment and y-axis labelling of figure 2 panel C; the second (2021;181(6):883) corrects a minor discrepancy between the study protocol and the description of how messages were conveyed to participants during the trial, altering only the methods section. Neither the between-group −0.26 kg nor the appendicular lean mass index of −0.16 kg/m² cited here falls within the scope of those errata.

---

#Sources

• Garegnani LI, Oltra G, Ivaldi D, et al. Intermittent fasting for adults with overweight or obesity. Cochrane Database Syst Rev. 2026 Feb 16;2(2):CD015610 (doi:10.1002/14651858.CD015610.pub2; PMID 41692034) — inclusion criteria (a minimum intervention of four weeks, a minimum follow-up of six months, cross-over and quasi-randomised designs excluded), MD −0.33 against regular dietary advice (21 studies, 1,430 people), MD −3.42 against no intervention (6 studies, 427 people), the adverse event risk ratios, 22 trials and 1,995 people in all, no study reporting diabetes status, and the limit of applicability beyond 12 months. Traditional Chinese plain-language summary: https://www.cochrane.org/zh-hant/evidence/CD015610_intermittent-fasting-adults-overweight-or-obesity (translator Lin Yun-ting; the page notes that the English original prevails, so its Chinese sentences are not quoted here). Retrieved 3 August 2026.
• Parkinson M. Evidence behind intermittent fasting for weight loss fails to match hype. Cochrane (news), 16 February 2026: https://www.cochrane.org/about-us/news/evidence-behind-intermittent-fasting-weight-loss-fails-match-hype — the release's wording combining the two control arms, the inconsistent reporting of side effects, and the limit on generalising from a study population predominantly white and in high-income countries. Retrieved 2 August 2026.
• Semnani-Azad Z, Khan TA, Chiavaroli L, … Sievenpiper JL. Intermittent fasting strategies and their effects on body weight and other cardiometabolic risk factors: systematic review and network meta-analysis of randomised clinical trials. BMJ. 2025 Jun 18;389:e082007 (doi:10.1136/bmj-2024-082007; PMID 40533200; PMC12175170) — 99 trials and 6,582 people, alternate-day fasting against continuous energy restriction at −1.29 kg, the results by 24-week stratum, the minimally important difference of 2.0 kg, only five trials of 52 weeks or longer, no difference for glycated haemoglobin or high density lipoprotein, and the account of adverse events. The erratum BMJ. 2025 Aug 18;390:r1737 (PMID 40825602) corrects only one co-author's affiliation and no result. Retrieved 2 August 2026.
• Wang B, Wang C, Li H. The impact of intermittent fasting on body composition and cardiometabolic outcomes in overweight and obese adults: a systematic review and meta-analysis of randomized controlled trials. Nutrition Journal, published 30 July 2025 (doi:10.1186/s12937-025-01178-6; PROSPERO CRD420251036588): https://link.springer.com/article/10.1186/s12937-025-01178-6 — 15 trials and 758 people, body weight −3.73 kg, triglycerides +13.22 mg/dl, total cholesterol, diastolic blood pressure, and the inclusion criterion excluding studies with a calorie-restriction control group. Retrieved 2 August 2026.
• Liu D, Huang Y, Huang C, et al. Calorie Restriction with or without Time-Restricted Eating in Weight Loss. N Engl J Med. 2022 Apr 21;386(16):1495-1504 (doi:10.1056/NEJMoa2114833; PMID 35443107; NCT03745612) — 12 months, 139 people, a net difference of −1.8 kg (P = 0.11), completion 84.9%. Retrieved 2 August 2026.
• Trepanowski JF, Kroeger CM, Barnosky A, et al. Effect of Alternate-Day Fasting on Weight Loss, Weight Maintenance, and Cardioprotection Among Metabolically Healthy Obese Adults. JAMA Intern Med. 2017 Jul 1;177(7):930-938 (doi:10.1001/jamainternmed.2017.0936; PMID 28459931; NCT00960505) — the one-year design, dropout of 38%/29%/26% and the authors' own account of selection bias and reduced power, adherence, LDL showing no difference at month 6 and +11.5 mg/dL at month 12, total cholesterol showing no difference at either time point, the HDL results at months 6 and 12, and the authors' explanation that participants were primarily metabolically healthy obese adults. The paper records that one of its authors received an advance from a publisher for a book on alternate-day fasting. Retrieved 2 August 2026.
• Lowe DA, Wu N, Rohdin-Bibby L, et al. 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. 2020 Nov 1;180(11):1491-1499 (doi:10.1001/jamainternmed.2020.4153; PMID 32986097) — 12 weeks, 116 people, a between-group −0.26 kg, an appendicular lean mass index of −0.16 kg/m² (in-person visit cohort of 50 people). There are two errata (same journal, 2020;180(11):1555, PMID 33136126, correcting figure 2 panel C; and 2021;181(6):883, PMID 33616618, correcting the methods description of how messages were conveyed to participants); the content of both has been obtained and neither affects the figures cited here. The discussion of that paper is also the original source of the "about 65% of the weight lost was lean mass" figure, and its limitations state that protein intake was not measured and that dual-energy X-ray absorptiometry does not account for muscle hydration. Retrieved 3 August 2026.
• Jamshed H, Steger FL, Bryan DR, et al. Effectiveness of Early Time-Restricted Eating for Weight Loss, Fat Loss, and Cardiometabolic Health in Adults With Obesity. JAMA Intern Med. 2022 Sep 1;182(9):953-962 (doi:10.1001/jamainternmed.2022.3050; PMID 35939311) — 14 weeks, 90 people, a 07:00–15:00 window, a control group with a window of 12 hours or more a day, weight −2.3 kg, body fat showing no difference in the randomised analysis but a difference in the prespecified completers analysis (59 people), diastolic blood pressure −4 mm Hg, and the equivalent of 214 kcal a day. Retrieved 3 August 2026.
• Schroor MM, Joris PJ, Plat J, Mensink RP. Effects of Intermittent Energy Restriction Compared with Those of Continuous Energy Restriction on Body Composition and Cardiometabolic Risk Markers. Adv Nutr. 2024 Jan;15(1):100130 (doi:10.1016/j.advnut.2023.10.003; PMID 37827491) — 28 included studies (26 original trials), fat-free mass −0.20 kg, waist circumference −0.91 cm, no difference in BMI, glucose, insulin, lipids or blood pressure, alternate-day fasting improving the insulin resistance index more, and the 5:2 diet reducing BMI less. Retrieved 3 August 2026.
• Guo L, Xi Y, Jin W, et al. A 5:2 Intermittent Fasting Meal Replacement Diet and Glycemic Control for Adults With Diabetes: The EARLY Randomized Clinical Trial. JAMA Netw Open. 2024 Jun 3;7(6):e2416786 (doi:10.1001/jamanetworkopen.2024.16786; PMID 38904963; ChiCTR2000040656) — 16 weeks, 405 randomised, 332 completing, the glycated haemoglobin and weight results, and the authors' framing of it as an "initial" intervention. Retrieved 2 August 2026.
• Chen M, Xu L, Van Horn L, Manson JE, Tucker KL, Du X, Feng N, Rong S, Zhong VW. Association of eating duration less than 8 h with all-cause, cardiovascular, and cancer mortality. Diabetes Metab Syndr. 2025 Jul;19(7):103278 (doi:10.1016/j.dsx.2025.103278; PMID 40849219) — n = 19,831, median follow-up 8.1 years, a cardiovascular mortality hazard ratio of 2.35 (95% CI 1.39–3.98), that association holding across 8 subgroups and 14 sensitivity analyses, the all-cause association failing many sensitivity analyses, and the authors' reservation about residual confounding. Retrieved 3 August 2026.
• American Heart Association Newsroom. 8-hour time-restricted eating linked to a 91% higher risk of cardiovascular death. Published 18 March 2024: https://newsroom.heart.org/news/8-hour-time-restricted-eating-linked-to-a-91-higher-risk-of-cardiovascular-death — the content of conference abstract P192 of 2024, the AHA's own statement that abstracts are not peer reviewed, the senior author's reservation about causation, and all four points of Dr Gardner's independent comment. Retrieved 3 August 2026.
• American Diabetes Association Professional Practice Committee. 5. Facilitating Positive Health Behaviors and Well-being to Improve Health Outcomes: Standards of Care in Diabetes—2026. Diabetes Care. 2026 Jan 1;49(Suppl 1):S89-S131 (doi:10.2337/dc26-S005; PMID 41358898; PMC12690188) — the three-way definition of intermittent fasting, the nonreligious fasting passage (including the 3–8% loss over 8–12 weeks and no significant difference against continuous calorie restriction, and the waist circumference and fat-free mass results of the 2024 pooled analysis), Figure 5.1 on the similarities and differences between religious and nonreligious fasting (three shared items, four differing items each, and the dehydration distinction), the statement on safety in diabetes, the four risks, the ranking of hypoglycaemia risk by drug class, the last few hours of a fast accounting for approximately 50% of hypoglycaemic events, the >300 mg/dL proportion, recommendations 5.32 and 5.33, the statements on eating disorders, and the relative ranking of the risk elements in table 5.3 and the hypoglycaemia risk column of table 5.4 (this article reproduces none of the scores, thresholds, risk levels or dose adjustments in either table). The full text (including tables 5.3 and 5.4 and figure 5.1) was read from the PMC full-text version: https://pmc.ncbi.nlm.nih.gov/articles/PMC12690188/. Retrieved 3 August 2026.
• Diabetes and Ramadan | Fasting | Diabetes UK: https://www.diabetes.org.uk/about-diabetes/looking-after-diabetes/ramadan — the exemption list, the list of what risk depends on, the instructions for handling low and high blood glucose, premixed insulin not being recommended, and the quotation from Professor Wasim Hanif. Page structured data: datePublished 21 August 2017, dateModified 1 July 2026 (no date displayed on the page). Retrieved 3 August 2026.
• Religious fasting and diabetes | Enjoy food | Diabetes UK: https://www.diabetes.org.uk/living-with-diabetes/eating/fasting — the explicitly illustrative exemption list, and the advice against fasting when unwell. Page structured data: datePublished 21 August 2017, dateModified 24 March 2026 (no date displayed on the page). Retrieved 3 August 2026.
• Varady KA, Cienfuegos S, Ezpeleta M, Gabel K. Clinical application of intermittent fasting for weight loss: progress and future directions. Nat Rev Endocrinol. 2022 May;18(5):309-321 (doi:10.1038/s41574-022-00638-x; PMID 35194176) — an alternative definition of alternate-day fasting (0–500 kcal on fast days; the original wording is "the most studied forms"). Competing interests: K.A.V. received an author fee from Hachette Book Group for The Every Other Day Diet. Retrieved 3 August 2026.
• Titmus M. The influence of time-restricted eating on weight management and metabolic health. British Dietetic Association, 28 September 2021: https://www.bda.uk.com/resource/the-influence-of-time-restricted-eating-on-weight-management-and-metabolic-health.html — the description of mild adverse events and the adherence note that follows, the behavioural signal and its self-resolving outcome (the underlying material being Bjerre et al.'s qualitative interviews with 17 participants), and the author's own reservation about the muscle loss question. This is a signed review article, not a BDA position statement. Retrieved 2 August 2026.
• Centre for Health Protection "Obesity" health topic page (page dated 26 April 2023): Chinese https://www.chp.gov.hk/tc/healthtopics/content/25/8802.html / English https://www.chp.gov.hk/en/healthtopics/content/25/8802.html — obesity at 32.6% and overweight at 22.0% among people aged 15–84, the body mass index cut-offs for Chinese adults in Hong Kong, and the centre's single sentence of advice on weight management. Retrieved 2 August 2026.
• Department of Health, Population Health Survey 2020-22, part 2 report (Chinese version https://www.chp.gov.hk/files/pdf/dh_phs_2020-22_part_2_report_chi_rectified.pdf; English version https://www.chp.gov.hk/files/pdf/dh_phs_2020-22_part_2_report_eng_rectified.pdf; topic page dated 11 November 2025) — the age-standardised prevalence for people aged 18–84 (51.3% on the Asian classification, 30.8% on the WHO one), the survey's coverage and the 2,072 people who completed a physical examination, the 55.2% participation rate, the weight self-perception figures, and the 43.3% with undiagnosed overweight or obesity. Retrieved 2 August 2026.
• Centre for Health Protection "Physical Activity" health topic page (page dated 15 April 2025): Chinese https://www.chp.gov.hk/tc/healthtopics/content/100200/8804.html / English https://www.chp.gov.hk/en/healthtopics/content/100200/8804.html — the World Health Organization recommendation of 150–300 minutes of moderate-intensity aerobic activity a week and muscle strengthening on two days a week (quoted by the Centre for Health Protection), and the 14.8% found insufficiently active in the Behavioural Risk Factor Survey of 2023. Retrieved 2 August 2026.
• Hong Kong Dietitians Association, "Dietitians on Media" subject index: https://www.hkda.com.hk/posts/176 (healthy eating and food nutrition), https://www.hkda.com.hk/posts/178 (obesity and the three highs) and ten other subject headings, together with its sitemap https://www.hkda.com.hk/p/sitemap/37561 — 533 records across the twelve subject headings, three intermittent fasting records and the external links they point to, and the absence from the sitemap of any position statement or guideline section. Retrieved 3 August 2026.
• Centre for Health Protection "Healthy Eating" resources page: https://www.chp.gov.hk/tc/resources/100201/8800.html — the list of 32 healthy eating publications (the page carries 35 links, of which 3 are duplicate accessible versions). The booklets 「你是否過胖」 https://www.chp.gov.hk/files/pdf/exn_nutp_007b_tc.pdf and 「熱量知多少」 https://www.chp.gov.hk/files/pdf/exn_nutp_058b.pdf and the 「中央肥胖」 poster https://www.chp.gov.hk/files/her/exn_nutp_064cp.pdf were each opened individually (the poster is an image file and was read as an image). Retrieved 3 August 2026.
• Change4Health, Department of Health: healthy eating "Guidelines" https://www.change4health.gov.hk/tc/healthy_diet/guidelines/index.html, "Facts" https://www.change4health.gov.hk/tc/healthy_diet/facts/index.html, FAQs https://www.change4health.gov.hk/tc/faqs/healthy_diet/index.html, and related links (local) https://www.change4health.gov.hk/tc/link/local/index.html — the list of four healthy eating guidelines, the healthy eating FAQs, and the official signposting to the Hong Kong Dietitians Association and the Hong Kong Nutrition Association. Retrieved 3 August 2026.

#Further reading

• On this platform: Weight-loss drugs

This article is health information, not medical advice. Any personal decision about fasting, diet or medication should be discussed with your doctor or medical team.