None of this is a personal failing. Cravings are a normal part of how brains work, sweeteners are a topic where even expert bodies disagree, and labels differ from country to country in ways that are rarely explained. This chapter helps you make sense of all three without guilt.

We begin with cravings: where they come from, what may calm them, and whether the order in which you eat a meal matters. We then look at sugar, sweeteners, and "natural" alternatives, including what the World Health Organization (WHO) says and, just as important, what it does not say. We finish with a practical guide to labels, including how to work out the carbohydrate in a portion from a "per 100 g" panel. Chapters 4 to 7 gave you the background, and we build on them here.

8.1 Understanding Cravings

What a craving is, and what it is not

A craving is an intense desire for a specific food, usually a sweet, fatty, or salty one. It is not the same as hunger. Reviews describe it as a conditioned response triggered by the sight, smell, or even the thought of a food, and involving the brain's reward circuitry, which is driven by the chemical messenger dopamine. The same review points out that people often eat when there is no metabolic signal that the body needs fuel.[@rebello2016] So a craving is a learned reaction to a cue, not a verdict on your character.

People who find it hard to stick to strict energy restriction also report more cravings.[@rebello2016] That is an association and may run in either direction, but it is one reason to be wary of rigid "never again" rules about favourite foods.

Hunger is not only about glucose

There is something in the idea that a glucose "crash" after a meal drives hunger. In a study using continuous glucose monitors in 1,070 people, bigger dips two to three hours after a meal were associated with more hunger, a shorter wait until the next meal, and more energy eaten later. The associations were small (a correlation of 0.16 for hunger), and the participants were healthy people without diabetes.[@wyatt2021] If you take insulin or medicines that can cause low glucose and feel sudden hunger with shakiness, sweating, or confusion, treat it as a possible low, check your glucose (Chapter 4), and follow your care team's plan.

Sleep and stress

A meta-analysis of studies that deliberately shortened sleep found that people ate about 385 kilocalories more after short sleep (95 percent confidence interval 252 to 517), with no matching rise in energy expenditure. These were mostly short laboratory studies in healthy adults (11 studies, 172 participants), so real life may differ.[@alkhatib2017] A meta-analysis of 54 studies with 119,820 participants linked stress to a small increase in unhealthy food intake and a small decrease in healthy intake, with very large differences between people.[@hill2022] Chapter 10 covers sleep and stress; the point here is that cravings that peak after a bad night or a hard week are common and understandable.

What tends to help

The evidence for any single trick is modest, but several practical steps are reasonable.

Build meals that keep you full. Chapter 6 showed how vegetables, protein, and unrefined carbohydrate fill the plate. A systematic review of randomised trials found that more viscous (gel-forming) fibres, such as pectins, beta-glucans, and guar gum, reduced appetite in 59 percent of comparisons, against 14 percent for less viscous fibres, although the overall effects on energy intake and body weight were modest.[@wanders2011] Oats, barley, pulses, and many fruits are rich in these fibres. Pulses are discussed in Chapter 7.

Change the cue. Because cravings are triggered by sensory cues,[@rebello2016] it can help to keep the foods that call to you out of sight and out of the house, and to put fruit, nuts, or yoghurt where your eye lands first. This is a practical suggestion rather than a proven treatment.

Plan the sweet thing instead of forbidding it. Choosing when and how much, for instance a small portion with a meal, removes the sense of secrecy and guilt.

Try mindful eating, with realistic expectations. In a pilot trial in adults with type 2 diabetes, 52 people who completed either a three-month mindful-eating programme or standard diabetes education both lowered their HbA1c (by 0.83 and 0.67 percentage points respectively), with no significant difference between the two.[@miller2012] A review of 31 trials of mindfulness-based programmes in type 2 diabetes reported a small average HbA1c reduction of 0.44 percentage points, with low certainty, effects that faded by two to three years, and only about three of the studies focusing on mindful eating itself.[@ee2025] Mindfulness may help you notice what a craving feels like, but it is not a proven glucose treatment.

If food feels out of control, if you often eat large amounts in a short time and feel distressed afterwards, or if thoughts about food take up much of your day, please tell your doctor or a mental-health professional. This is common, it is treatable, and it is not a moral weakness. Chapter 16 returns to emotional wellbeing.

The order you eat in

Because the same meal can be eaten in different orders, researchers have asked whether eating vegetables and protein first and carbohydrate last changes the glucose response. The best-known study was small: 11 overweight or obese adults with metformin-treated type 2 diabetes ate the same 628-kilocalorie meal (68 g carbohydrate) in two orders, a week apart. When vegetables and protein came first, glucose was 28.6 percent lower at 30 minutes, 36.7 percent lower at 60 minutes, and 16.8 percent lower at 120 minutes, and insulin was lower too.[@shukla2015] A systematic review of six short studies in 107 healthy adults also found lower glucose in the first 15 to 45 minutes when vegetables, fruit, or protein preceded carbohydrate.[@kim2026]

Longer-term evidence is thinner. A 16-week pilot in 45 adults with prediabetes (39 completed) found a small HbA1c fall with carbohydrate-last counselling (-0.1 percentage points, against -0.03 in controls, not statistically significant) and no significant change in glucose tolerance.[@shukla2023] So food order is a free habit that may soften the after-meal rise in the short term. It is not a substitute for portion size, nor a proven route to better HbA1c. If you use mealtime insulin, a change in the order or timing of your meal is something to discuss with your care team first.

Science Corner: Where does "up to 70 percent" come from? Popular articles sometimes claim that eating in the right order cuts glucose spikes "by up to 70 percent". The nearest source appears to be the 11-person study above, in which one summary measure, the incremental area under the two-hour glucose curve, was 73 percent lower.[@shukla2015] That is one measure in 11 people eating one supervised meal. The reductions at individual time points were about 17 to 37 percent, and the longer pilot found only small HbA1c changes.[@shukla2023] The fair summary is "modest to moderate, in small studies".

8.2 Sugar, Sweeteners, and "Natural" Alternatives

Sugar in its many forms

The WHO uses the term free sugars for sugars added to foods and drinks by the manufacturer, cook, or consumer, together with the sugars naturally present in honey, syrups, fruit juices, and juice concentrates. It excludes the sugars inside whole fresh fruit, vegetables, and milk. Its strong recommendation is to keep free sugars below 10 percent of total energy, with a conditional suggestion that going below 5 percent, roughly 25 grams a day, would give additional benefit.[@who2015sugars] Sugar-sweetened drinks are the easiest place to start (Chapters 3 and 6). European diabetes guidance rates as high-certainty the evidence that high sugar intake, especially from sugary drinks at 10 percent of energy or more, increases body weight.[@easd2023]

Non-sugar sweeteners: what the WHO says, and to whom

The WHO's 2023 guideline is often quoted without its details, so here they are. Its wording is: "WHO suggests that non-sugar sweeteners not be used as a means of achieving weight control or reducing the risk of noncommunicable diseases", and it is labelled a conditional recommendation based on low certainty evidence overall. The randomised trials it reviewed averaged three months or less, and the cohort studies followed people for about 13 years. Short trials showed modest weight loss when these sweeteners replaced free sugars, and cohort studies linked higher intakes with higher risks of type 2 diabetes, cardiovascular disease, and death, but observational findings cannot separate the sweetener from other differences between people.[@who2023sweet]

Two exclusions matter for readers of this book. First, the recommendation does not apply to people with pre-existing diabetes, because studies in that group were excluded from the review. Second, low-calorie sugars and sugar alcohols (polyols) are not counted as non-sugar sweeteners in the guideline, so it does not cover them.[@who2023sweet] The sweeteners it covers include acesulfame K, aspartame, saccharin, sucralose, and stevia and its derivatives.[@who2023news]

Other expert bodies read the same field differently. The EASD's 2023 nutrition recommendations state that low-calorie-sweetened beverages, when replacing sugar-sweetened beverages, reduce body weight and cardiometabolic risk factors, rating the evidence as moderate to high certainty.[@easd2023] A meta-analysis of 29 randomised trials (741 participants, some with diabetes) found that non-nutritive sweeteners did not raise blood glucose,[@nichol2018] although its authors called for studies of long-term use.

These positions are less contradictory than they look. One asks whether sweeteners protect the general population from weight gain and disease in the long run (the WHO); the other asks whether a swap is better than the sugary drink itself (the EASD). A cautious reading is that water and unsweetened drinks are the best default, and that a swap from a sugary drink is a reasonable step towards it, not a reason to sweeten everything. The WHO itself suggests cutting free sugars by choosing fruit or unsweetened foods and drinks.[@who2023news]

Sugar alcohols and erythritol

Sugar alcohols (polyols) include erythritol, xylitol, sorbitol, and maltitol. Despite the name they are neither sugars nor alcohols, and they provide about half the calories of sugar.[@adacarbs] They differ: an expert position paper gives an average of about 2 kilocalories per gram, from 0.2 for erythritol to 4.0 for isomaltulose, and notes that doses around 30 grams a day of some polyols can cause gas, cramps, and loose stools.[@academy2012] European labelling rules likewise assign polyols in general 2.4 kilocalories per gram, and erythritol 0.[@eu1169] Because they have been used for years to cut the carbohydrate that raises glucose, many "sugar-free" and "no added sugar" products contain them.[@academy2012]

Erythritol has received particular attention since 2023. A study in Nature Medicine measured blood erythritol in people undergoing cardiac evaluation: 1,157 in a discovery group, 2,149 in a US group, and 833 in a European group. Those in the top quarter for erythritol had about twice the rate of major cardiovascular events over three years as those in the bottom quarter, with hazard ratios of 1.80 (95 percent confidence interval 1.18 to 2.77) in the US group and 2.21 (1.20 to 4.07) in the European group. In the laboratory, erythritol made platelets more reactive, and in eight healthy volunteers an erythritol-sweetened drink raised blood levels more than 1,000-fold for at least two days.[@witkowski2023][@nih2023eryth] The authors described an association and called for long-term safety studies.[@witkowski2023]

Science Corner: Why erythritol is a caution, not a verdict The study measured erythritol in blood, not in the diet. Critics point out that the body makes some erythritol itself from glucose, that people with the highest levels were on average 10 to 14 years older and more often had diabetes, and that no dietary information was collected before sampling; they conclude that no causal relationship can be derived.[@cramer2023] The eight-person test showed that blood levels rise, not that clots or heart attacks follow. A sensible reading is that the signal is not proof, deserves follow-up, and is one more reason not to make heavily sweetened products of any kind a daily staple. If you already have heart disease or high cardiovascular risk, ask your care team about regular erythritol use.

Allulose, stevia, and monk fruit

These are categories, and the evidence is limited. Allulose is a "rare sugar"; in a meta-analysis of six small, short trials in 126 people with type 2 diabetes it lowered the after-meal glucose response.[@ayesh2024] Stevia is among the WHO's non-sugar sweeteners, and monk fruit (luo han guo) extract is among those approved in the United States.[@who2023news][@academy2012] This chapter cites no long-term trials in people with diabetes for either. The label matters more than the marketing.

Honey, jaggery, palm sugar, and agave

It is tempting to assume that "natural" sweeteners are kinder to blood glucose. The evidence does not support treating them as a licence. The WHO counts honey and syrups as free sugars.[@who2015sugars] A systematic review of eight randomised trials of honey in people with diabetes found mixed results: seven reported reductions in glucose or HbA1c, two no difference, and two increases, and the authors concluded that no definite conclusion could be drawn.[@zamanian2020] In a randomised crossover study of 43 adults with type 2 diabetes, coconut jaggery and cane sugar (50 g of digestible carbohydrate each) produced comparable glucose responses, and the authors concluded that jaggery cannot be considered a healthy substitute.[@thilanka2022] Agave syrup has a low glycemic index, but is mostly fructose, has more calories and grams of sugar than table sugar, and, as one university extension service puts it, "is still sugar; it will still raise blood glucose levels".[@illinois2018] This chapter did not find trial evidence for palm sugar or date syrup, so treat them the same way. A lower glycemic index does not make a sweetener free, since it still adds carbohydrate and free sugars (Chapter 5).

8.3 Reading Labels Worldwide

For glucose, the first question about any packaged food is: how many grams of carbohydrate will I actually eat? Chapter 7 showed how to estimate this from food tables; here we use the label.

Total carbohydrate: US and EU conventions differ

In the United States, the ADA explains that "total carbohydrate" on the Nutrition Facts label covers starch, fibre, and sugar together.[@adacarbs] In the European Union and the UK, the legal definition of carbohydrate is any carbohydrate metabolised by humans, including polyols, and fibre is a separate, optional line.[@eu1169] A laboratory-industry presentation states that the EU carbohydrate figure does not include fibre.[@eurofins2025] So the same wholemeal bread can show a larger carbohydrate number on a US label than on a European one, purely because of how it is counted. Other countries follow their own rules, so learn how your own country's label works and compare like with like; a dietitian or your national food-safety agency can tell you.

"Net carbs" is not a regulated term

"Net carbs" usually means total carbohydrate minus fibre and sugar alcohols. The ADA states that the term has no legal definition, is not used by the US Food and Drug Administration (FDA), and is not recognised by the ADA, and that the fibre and sugar-alcohol contribution depends on the specific types, which labels do not give.[@adacarbs][@adafoodhub2026] The ADA advises using total grams of carbohydrate. Rules elsewhere differ, so treat "net carbs" on any pack as a marketing calculation.

Science Corner: How "net" can go wrong Imagine a 40 g bar whose EU-style label reads carbohydrate 22 g (of which polyols 14 g), fibre 8 g. Subtracting the polyols gives 8 g. Subtracting the fibre as well gives 0 g, yet fibre was never part of the 22 g in that labelling system, so it has been removed twice. (These numbers are invented and were calculated by script.) Polyols also differ from one another in how much they raise glucose, so even the first subtraction is only an estimate.[@academy2012] If you count carbohydrate, or use mealtime insulin, ask your care team which method they want you to use.

Sugars, added sugars, and free sugars

The "of which sugars" line is total sugars, natural (in milk and fruit) and added together. In the US, a separate "Added sugars" line covers sugars added in processing, syrups, honey, and concentrated juices, with a Daily Value of 50 grams a day on a 2,000-kilocalorie diet.[@fdaadded] In the UK, the NHS lists per-100 g thresholds for total sugars: 5 g or less is low, and more than 22.5 g is high.[@nhslabels] The WHO's free sugars (section 8.2) are a third way of counting. For glucose, all of these sugars are already inside the total carbohydrate you count; the sugar lines show mostly how refined a food is.

Fibre

In the US system fibre sits inside total carbohydrate. The ADA advises people with diabetes to eat at least 14 grams per 1,000 kilocalories,[@adacarbs] the WHO gives at least 25 grams a day,[@whodiet] and EASD at least 35 grams.[@easd2023] More fibre per 100 g usually means a less refined food (Chapters 5 to 7).

Serving size versus per 100 g

This is where many mistakes start. Values on a US label refer to one serving, and if you eat two servings you double everything.[@fdalabel] In the EU, values are given per 100 g or 100 ml, with per-portion figures as an optional addition;[@eu1169] UK labels show per 100 g and sometimes per portion.[@nhslabels] The "serving" is set by the manufacturer, not by your appetite. The calculation is:

Carbohydrate in your portion (g) = carbohydrate per 100 g × grams you eat ÷ 100

The table applies it to five made-up products. The carbohydrate, sugar, and fibre values are invented for illustration, and the portion columns were calculated by script.

Illustrative label values (not real products); the portion columns were calculated as value per 100 g × portion ÷ 100. Note how similar the carbohydrate is across very different foods, while sugars and fibre differ sharply.

Two lessons follow. First, very different foods can deliver about the same carbohydrate per portion, which is why the amount matters more than the look of the pack (Chapters 5 and 6). Second, the sugar arrives very differently: the fruit drink is almost all sugar, the flatbread almost none. The same arithmetic gives the weight for 30 grams of carbohydrate, as in Chapter 7: 30 ÷ 62 × 100 is about 48 g of the cereal.

"Sugar-free", "no added sugar", and other claims

These phrases have legal meanings that vary by country. Under the UK-retained EU claims rules, "sugars-free" needs no more than 0.5 g of sugars per 100 g or 100 ml, and "low sugars" no more than 5 g per 100 g (2.5 g per 100 ml for liquids). "With no added sugars" means no added mono- or disaccharides or other foods used for sweetening, and the pack should say if it contains naturally occurring sugars.[@eu1924] The US also defines these claims, but the ADA cautions that "sugar-free" does not always mean healthy, and that low-carb, keto, and paleo claims have no FDA rules.[@adafoodhub2026] A "sugar-free" biscuit still contains starch, and may contain polyols or fat. Check total carbohydrate.

Front-of-pack symbols

Many countries add a summary symbol. UK traffic-light labels show fat, saturated fat, sugars, and salt in red, amber, or green.[@nhslabels] Nutri-Score, used voluntarily in seven European countries, grades products from A to E per 100 g, crediting fibre, protein, fruit, vegetables, and pulses, and penalising energy, saturated fat, sugars, salt, and non-nutritive sweeteners.[@nutriscore] Chile requires black octagon warnings for products "high in" sugars, sodium, saturated fat, or calories; a before-and-after study found a 23.7 percent fall in purchases of "high-in" beverages, though that design cannot prove the labels alone were responsible.[@taillie2020]

These symbols are built to rate overall healthiness for the general public. They do not tell you the grams of carbohydrate in your portion, and a well-rated cereal can still contain 28 grams in a normal bowl (see table). Use them as a quick screen, then read the numbers.

The ingredient list

Ingredients are listed from the largest amount to the smallest, so sugar in the first three lines means it is a major ingredient. Sugar has many names: sucrose and dextrose, but also syrups, honey, and concentrated fruit or vegetable juices, all of which count as added sugars.[@fdaadded] A product can contain several types, none of them first, that together add up.

{{fig:F8-A}}

Key Takeaways

• A craving is a learned reaction to a cue and the brain's reward system, not a character flaw. Strict rules can make cravings louder, so plan sweet foods instead of banning them.
• Poor sleep and stress are linked, on average and with wide variation, to eating more of the foods people crave (Chapter 10).
• Eating vegetables and protein before carbohydrate lowered after-meal glucose by roughly 17 to 37 percent at single time points in small short studies. "Up to 70 percent" overstates it, and long-term benefit is unproven.
• The WHO's 2023 advice on non-sugar sweeteners is conditional, rests on low-certainty evidence, and does not apply to people with existing diabetes. EASD guidance supports low-calorie-sweetened drinks as a swap for sugary ones. Water remains the best default.
• Erythritol was linked to cardiovascular events in an observational study plus a small laboratory test: a caution, not proof.
• Honey, jaggery, and agave are still sugars; "natural" or "low GI" does not make them free.
• On labels, check the serving size, use total carbohydrate for your portion, and be sceptical of "net carbs". Front-of-pack symbols do not show grams of carbohydrate.

Action Points

1. Notice, name, then decide. Next time a craving arrives, note the time, your sleep, and the cue. Notice patterns for a week before changing anything.
2. Try carbohydrate last at one meal a day, and check how you feel and, if you have a meter or sensor, your glucose (Chapters 4 and 12). If you take insulin, talk to your care team first.
3. Swap one sugary drink for water or an unsweetened option, and treat sweeteners as a stepping stone, not a target.
4. Read one label properly. Find the serving size, the total carbohydrate, the sugars, and the fibre, and calculate carbohydrate for your actual portion.
5. Ask your care team how they want you to treat sugar alcohols and "net carbs", and whether regular erythritol use is a concern for you.

This book is intended for education and does not replace personal medical advice. If you have diabetes or take glucose-lowering medication, please consult your healthcare team before changing your diet, exercise, or treatment.