Place · Level 3
Glycemic Index & Glycemic Load
GI 测的是实验室里 50 克碳水的平均升糖曲线 · GL 把份量乘进去 · 个体差异让人人反应不同 · 硬证据在糖尿病血糖管理, 不是减肥魔法
Story path
- 1GI is a lab curveGI is a lab curve
- 2GL multiplies in the portionGL multiplies in the portion
- 3Site of action: absorption rate → insulin curveSite of action: absorption rate → insulin curve
- 4Same food, different peopleSame food, different people
- 5Low-GI is not weight-loss magicLow-GI is not weight-loss magic
- 6In practice: a lens, not a lawIn practice: a lens, not a law
Chapter 1
GI is a lab curve
GI is a lab curve
You take a bite of rice; in the small intestine the starch is broken into glucose, absorbed into blood, and blood glucose starts to rise — the glycemic index (GI) measures how steep that rise is.
It's measured like this: a group of healthy, fasted volunteers each eat a food containing 50 g of available carbohydrate (roughly total carbs minus fiber); the area under their 2-hour blood-glucose curve is compared with the area after eating the same amount of pure glucose, with glucose set at 100. Jenkins and colleagues first proposed this method in 1981.
So a high GI tells you only one thing: at this standard portion, the food raises blood glucose fast and high. It doesn't tell you whether the food is healthy or whether you should eat it, and it isn't your personal response — because GI is the average of a small group of people, usually only 5 to 10 per group, not your curve.
It's measured like this: a group of healthy, fasted volunteers each eat a food containing 50 g of available carbohydrate (roughly total carbs minus fiber); the area under their 2-hour blood-glucose curve is compared with the area after eating the same amount of pure glucose, with glucose set at 100. Jenkins and colleagues first proposed this method in 1981.
So a high GI tells you only one thing: at this standard portion, the food raises blood glucose fast and high. It doesn't tell you whether the food is healthy or whether you should eat it, and it isn't your personal response — because GI is the average of a small group of people, usually only 5 to 10 per group, not your curve.
Method · why 50 g, why minus fiber
Why fix 50 g of available carb? To compare different foods fairly at the same sugar load, you hold the carb amount constant and let only the food's structure (particle size, fiber, fat, processing) shape the curve.Why available carb, minus fiber? Human enzymes can't cleave fiber, so it barely raises blood glucose directly (it travels to the colon to feed gut bacteria — see the `carbs-fiber` chapter). What actually becomes glucose in the small intestine is the starch and sugar.
Why the 2-hour area under the curve, not the peak height? A sharp high peak with a fast fall and a gentle low peak with a long tail can share the same peak yet mean different things for the body. The area captures both how high and how long.
Chapter 2
GL multiplies in the portion
GL multiplies in the portion
GI has a blind spot: it fixes the amount of carb (50 g) but ignores how much you actually eat. The total glucose reaching your blood depends on how much available carb the meal truly contains — that's the missing half that glycemic load (GL) supplies.
The formula is simple: GL = GI × the serving's actual available carb (g) ÷ 100. The GL concept was first introduced by Salmerón and colleagues in a 1997 cohort study of men.
The classic example is watermelon. It's often labeled high-GI, yet a normal slice (~120 g of flesh) contains very little glucose-raising carb, so a serving's GL is only about 4 (low). By contrast a bowl of white rice (GI ~73) reaches a GL of about 30 (high) per serving, while lentils (GI ~30) sit at a GL of about 8 (low).
So blanket 'avoid all high-GI foods' is too crude: what actually sets a meal's glucose exposure is GL — GI times the portion you really eat.
The formula is simple: GL = GI × the serving's actual available carb (g) ÷ 100. The GL concept was first introduced by Salmerón and colleagues in a 1997 cohort study of men.
The classic example is watermelon. It's often labeled high-GI, yet a normal slice (~120 g of flesh) contains very little glucose-raising carb, so a serving's GL is only about 4 (low). By contrast a bowl of white rice (GI ~73) reaches a GL of about 30 (high) per serving, while lentils (GI ~30) sit at a GL of about 8 (low).
So blanket 'avoid all high-GI foods' is too crude: what actually sets a meal's glucose exposure is GL — GI times the portion you really eat.
Detail · even GI values drift
Treating watermelon as high-GI is a textbook staple, but one thing is worth saying out loud: the GI of the same food drifts a lot between studies. Early single measurements reported watermelon at 72 to 76 (high), while the 2021 international GI/GL tables — pooling thousands of global data points — revised its average GI down to about 50 (low-moderate).That's not a contradiction; it confirms what GI really is: a measured group average that is inherently noisy — swayed by variety, ripeness, the test population, and blood-sampling method. So memorizing a food's GI to the single digit and treating it as law misreads the metric.
Only two conclusions are solid: first, multiplying in the portion (GL) is closer to reality than GI alone; second — as the next scene shows — what else is on the plate and who you are matter more than the number in the table.
Chapter 3
Site of action: absorption rate → insulin curve
Site of action: absorption rate → insulin curve
To see where a food's GI actually plays out in the body, watch two places: the small intestine and the pancreas.
In the small intestine, digestive enzymes break starch and sugar into glucose, which crosses the gut wall into blood. The faster it's cleaved and absorbed, the sharper and higher the glucose peak; slower cleavage makes a lower, flatter curve. The pancreas's β cells sense the rise and secrete insulin to move glucose into cells — the steeper the glucose surge, the bigger the insulin wave.
So GI is really an outward readout of the rate of glucose release in the small intestine. Anything that slows that release lowers the peak of the same portion of carbs:
Fat and protein in the same meal: slow gastric emptying and digestion. Jenkins's 1981 experiments already saw that more fat and protein meant a lower postprandial glucose rise.Fiber and viscosity: soluble fibers like oat β-glucan and legumes thicken gut contents, so glucose reaches the gut wall more slowly.Processing and particle size: finer-milled and softer-cooked starch is easier for enzymes to reach, raising glucose faster (intact grain vs flour).Resistant starch from cooling: when cooked rice or potato cools, some starch recrystallizes into hard-to-digest resistant starch, lowering the glycemic response.
In the small intestine, digestive enzymes break starch and sugar into glucose, which crosses the gut wall into blood. The faster it's cleaved and absorbed, the sharper and higher the glucose peak; slower cleavage makes a lower, flatter curve. The pancreas's β cells sense the rise and secrete insulin to move glucose into cells — the steeper the glucose surge, the bigger the insulin wave.
So GI is really an outward readout of the rate of glucose release in the small intestine. Anything that slows that release lowers the peak of the same portion of carbs:
Fat and protein in the same meal: slow gastric emptying and digestion. Jenkins's 1981 experiments already saw that more fat and protein meant a lower postprandial glucose rise.Fiber and viscosity: soluble fibers like oat β-glucan and legumes thicken gut contents, so glucose reaches the gut wall more slowly.Processing and particle size: finer-milled and softer-cooked starch is easier for enzymes to reach, raising glucose faster (intact grain vs flour).Resistant starch from cooling: when cooked rice or potato cools, some starch recrystallizes into hard-to-digest resistant starch, lowering the glycemic response.
Evidence · cooled rice really does blunt the rise
Cooling-and-recrystallizing isn't folklore — small controlled trials support it. In a 2015 randomized crossover study, researchers served white rice three ways: freshly cooked, cooled at room temperature for 10 hours, and refrigerated at 4°C for 24 hours then reheated. The refrigerated-then-reheated batch's resistant starch rose from 0.64 to 1.65 g per 100 g, and the postprandial glucose response of 15 healthy subjects was significantly lower than after freshly cooked rice.An honest note on effect size: this was a 15-person study; the difference is real but modest — it doesn't turn white rice into a low-GI staple. Its value is confirming the mechanism: the physical structure of starch (not merely whether it's a carb) sets how fast the small intestine breaks it down. To amplify the effect, rather than fussing over rice temperature, just switch to intact-grain, legume, and high-fiber carbs (covered in the final practice scene).
Chapter 4
Same food, different people
Same food, different people
We keep saying GI is an average. The strongest evidence for that comes from strapping glucose monitors onto many people and feeding them the same foods.
In 2015 the Weizmann Institute (Zeevi et al., in Cell) fitted 800 people with continuous glucose monitors (CGM) and logged nearly 47,000 real post-meal glucose responses. The result: for the same food, person-to-person differences in glucose response exceed food-to-food differences. One person spikes hard on a banana but stays flat on cookies; another is the reverse.
Where does the variation come from? The study traced it to a bundle of personal factors: gut microbiome composition, baseline insulin sensitivity, that day's sleep, the previous meal, recent exercise, and the time of day. A model trained on these features predicted a given person's response to a given meal fairly well — far better than the meal's GI alone.
This is why 'what's this food's GI?' is slowly giving way to 'how does my own body respond to it?' The meal's companions (fat, protein, fiber) and who you are together blur that GI table's authority.
In 2015 the Weizmann Institute (Zeevi et al., in Cell) fitted 800 people with continuous glucose monitors (CGM) and logged nearly 47,000 real post-meal glucose responses. The result: for the same food, person-to-person differences in glucose response exceed food-to-food differences. One person spikes hard on a banana but stays flat on cookies; another is the reverse.
Where does the variation come from? The study traced it to a bundle of personal factors: gut microbiome composition, baseline insulin sensitivity, that day's sleep, the previous meal, recent exercise, and the time of day. A model trained on these features predicted a given person's response to a given meal fairly well — far better than the meal's GI alone.
This is why 'what's this food's GI?' is slowly giving way to 'how does my own body respond to it?' The meal's companions (fat, protein, fiber) and who you are together blur that GI table's authority.
Don't overcorrect · small swings aren't the enemy
The mainstreaming of CGM brought a side effect: the anxiety that any glucose spike is bad. Time to tap the brakes.Healthy people's blood glucose is supposed to rise after a meal — that's the body working, not failing. A brief physiologic swing up and back after eating is not the same thing as the persistently elevated glucose of diabetes. Evidence indicates that, for people without diabetes, hunting down every small glucose peak as a threat is neither backed by hard evidence nor free of harm — it slides easily into over-restriction and food anxiety.
CGM genuinely helps in a narrower case: people who already have prediabetes, type 2 diabetes, or repeated post-meal symptoms can wear one for a week or two to find their own trigger foods (see the `cgm-for-healthy` and `type-2-diabetes` chapters). For a healthy person watching glucose out of pure curiosity, the payoff is limited and the cost is anxiety — that's worth saying honestly.
Chapter 5
Low-GI is not weight-loss magic
Low-GI is not weight-loss magic
Marketing often sells low-GI as a magic switch for fat loss and metabolism. Lay the evidence out and it's far less dramatic — and where the evidence is strong is not where the hype is.
The relatively solid part: glycemic management in type 2 diabetes. A 2009 Cochrane systematic review of 11 randomized controlled trials and 402 people with diabetes found that low-GI diets lowered the long-term glucose marker HbA1c by about 0.5 percentage points on average — modest but real, without more hypoglycemia. This is low-GI's most defensible use.
But even this is contested. The 2019 American Diabetes Association (ADA) nutrition consensus notes that systematic reviews of GI/GL disagree, and that studies define 'high' and 'low' GI so inconsistently that GI/GL's clinical usefulness is uncertain. Useful, in other words — but not the decisive lever.
The most-hyped, least-supported part: healthy people losing weight via low-GI. The 2023 updated Cochrane review (10 studies, 1,210 people) concluded there is insufficient evidence to draw any firm conclusion, with most studies small and certainty moderate-to-very-low. The 2018 DIETFITS randomized trial (609 people, 12-month follow-up) was blunter still: low-fat and low-carb groups lost similar amounts of weight, and a person's insulin-secretion level did not predict which diet suited them better — puncturing the claim that the carb-insulin axis is the master switch for weight loss.
The relatively solid part: glycemic management in type 2 diabetes. A 2009 Cochrane systematic review of 11 randomized controlled trials and 402 people with diabetes found that low-GI diets lowered the long-term glucose marker HbA1c by about 0.5 percentage points on average — modest but real, without more hypoglycemia. This is low-GI's most defensible use.
But even this is contested. The 2019 American Diabetes Association (ADA) nutrition consensus notes that systematic reviews of GI/GL disagree, and that studies define 'high' and 'low' GI so inconsistently that GI/GL's clinical usefulness is uncertain. Useful, in other words — but not the decisive lever.
The most-hyped, least-supported part: healthy people losing weight via low-GI. The 2023 updated Cochrane review (10 studies, 1,210 people) concluded there is insufficient evidence to draw any firm conclusion, with most studies small and certainty moderate-to-very-low. The 2018 DIETFITS randomized trial (609 people, 12-month follow-up) was blunter still: low-fat and low-carb groups lost similar amounts of weight, and a person's insulin-secretion level did not predict which diet suited them better — puncturing the claim that the carb-insulin axis is the master switch for weight loss.
Marketing claim vs the evidence
Claim: low-GI foods automatically burn fat and drop weight.Evidence: the 2023 Cochrane review says evidence is insufficient to conclude; DIETFITS says weight loss tracks adherence and overall diet quality, not a single carb-insulin axis. Calories and adherence remain the main line for weight loss.
Claim: low-GI is the key to reversing metabolism.
Evidence: in people with type 2 diabetes, low-GI modestly improves HbA1c (~0.5 points), worth using as an adjunct tool; but the ADA cautions that evidence is inconsistent and definitions are messy — it isn't a standalone deciding factor.
Why the gap? Because one phrase (low-GI) gets slid from the context of diabetic glucose management into the context of healthy-person weight loss. The former has controlled trials behind it; the latter is mostly short, small, wobbly studies. Treating the former's evidence as proof of the latter is the most common bait-and-switch.
An honest line: if you have diabetes or prediabetes, low-GI/GL is a handle worth discussing with your doctor or dietitian; if you're a healthy person just wanting to lose weight, what decides success is overall diet quality, calories, and adherence — not staring at a GI table.
Chapter 6
In practice: a lens, not a law
In practice: a lens, not a law
Pulling the five scenes into something usable, the core is: GI/GL is a lens for understanding food, not a list of good and evil foods.
First, look at the whole, don't chase single numbers. Evidence points consistently to this: what governs long-term metabolic health is the carbohydrate quality of the whole plate — more intact grains, legumes, and fiber-rich fruit and vegetables, less refined sugar and flour — not scoring every food by its GI. A low-GI sugary drink (spiked with fructose) doesn't become healthy just because its GI is low.
Second, use pairing and structure to flatten the curve, rather than quitting carbs. The same portion of carbs, paired with vegetables, protein, and good fats — or eaten after the veg and protein — lowers the glucose peak by slowing small-intestinal absorption (mechanism in scene 3). Intact beats milled; chewy beats cooked-to-mush.
Third, apply it to your situation. If you have type 2 diabetes or prediabetes, GI/GL plus CGM is a measured handle to use with your doctor and dietitian; if you're healthy, one normal-sized portion of a high-GI food inside a balanced meal is nothing to panic over.
Nutrition is a long game. This page is education, not a diagnosis, and does not replace individualized advice from a doctor or dietitian. But once you understand how glucose is released in the small intestine and why everyone differs, you're much harder to lead around by 'low-GI cures everything' marketing.
First, look at the whole, don't chase single numbers. Evidence points consistently to this: what governs long-term metabolic health is the carbohydrate quality of the whole plate — more intact grains, legumes, and fiber-rich fruit and vegetables, less refined sugar and flour — not scoring every food by its GI. A low-GI sugary drink (spiked with fructose) doesn't become healthy just because its GI is low.
Second, use pairing and structure to flatten the curve, rather than quitting carbs. The same portion of carbs, paired with vegetables, protein, and good fats — or eaten after the veg and protein — lowers the glucose peak by slowing small-intestinal absorption (mechanism in scene 3). Intact beats milled; chewy beats cooked-to-mush.
Third, apply it to your situation. If you have type 2 diabetes or prediabetes, GI/GL plus CGM is a measured handle to use with your doctor and dietitian; if you're healthy, one normal-sized portion of a high-GI food inside a balanced meal is nothing to panic over.
Nutrition is a long game. This page is education, not a diagnosis, and does not replace individualized advice from a doctor or dietitian. But once you understand how glucose is released in the small intestine and why everyone differs, you're much harder to lead around by 'low-GI cures everything' marketing.