Place · Level 3 · Debate
Energy Balance vs Carbohydrate-Insulin · An Ongoing Debate
营养学最有趣的当代争议 · CIM 立论 vs Hall 病房 RCT 反击 · 主流共识: 能量赤字决定, 宏量比例次要 · DIETFITS 头对头 12 个月几乎打平
Last updated
Story path
- 1The debate · two camps + why you should careThe debate · two camps + why you should care
- 2CIM hypothesis · a plausible mechanism storyCIM hypothesis · a plausible mechanism story
- 3Hall counterattack · metabolic-ward RCTs refute CIMHall counterattack · metabolic-ward RCTs refute CIM
- 4Mainstream consensus · what actually drives weight lossMainstream consensus · what actually drives weight loss
- 5Practical takeaway · don't get hijacked by extremesPractical takeaway · don't get hijacked by extremes
Chapter 1
The debate · two camps + why you should care
The debate · two camps + why you should care
The hottest academic debate in nutrition over the past decade boils down to a deceptively simple question: why do people get fat? Are they eating too much, or eating the wrong things?
Two camps:
Energy Balance Model (EBM) — mainstream consensus. Long-term intake > expenditure → weight gain; macronutrient ratios (carb/fat/protein) are secondary, influencing intake indirectly via satiety + adherence. Lead figures: Kevin Hall (NIH), Dale Schoeller.
Carbohydrate-Insulin Model (CIM) — proposed by Harvard's David Ludwig & Cara Ebbeling (Ludwig 2018). The causal arrow is reversed: high-GI carbs → postprandial insulin spike → surging fat storage → glucose nadir + hunger → overeating → obesity. Corollary: carbohydrate restriction / ketogenic diets are the right strategy for fat loss, and counting calories alone is futile.
Why ordinary people should care:
Influencer marketing has almost universally bet on CIM: 'carbs make you fat', 'insulin is the master switch of obesity', 'forget calories, just cut sugar', 'keto cures everything' — all far sexier than 'eat less, move more', and far easier to package into books, courses, and low-carb products.If CIM is right: decades of mainstream public-health advice (reduce total calories / eat a balanced diet) were wrong, and government dietary guidelines need rewriting.If CIM is wrong: you have been sold a simplified causal story and missed the actually useful levers (total energy + protein + food quality + sustainability).
Why the debate matters: not to pick a side, but to see the chasm between a mechanism story and clinical evidence. A mechanism that sounds airtight on paper can completely collapse inside a metabolic ward. Over the next three scenes we walk the evidence chain.
Two camps:
Energy Balance Model (EBM) — mainstream consensus. Long-term intake > expenditure → weight gain; macronutrient ratios (carb/fat/protein) are secondary, influencing intake indirectly via satiety + adherence. Lead figures: Kevin Hall (NIH), Dale Schoeller.
Carbohydrate-Insulin Model (CIM) — proposed by Harvard's David Ludwig & Cara Ebbeling (Ludwig 2018). The causal arrow is reversed: high-GI carbs → postprandial insulin spike → surging fat storage → glucose nadir + hunger → overeating → obesity. Corollary: carbohydrate restriction / ketogenic diets are the right strategy for fat loss, and counting calories alone is futile.
Why ordinary people should care:
Influencer marketing has almost universally bet on CIM: 'carbs make you fat', 'insulin is the master switch of obesity', 'forget calories, just cut sugar', 'keto cures everything' — all far sexier than 'eat less, move more', and far easier to package into books, courses, and low-carb products.If CIM is right: decades of mainstream public-health advice (reduce total calories / eat a balanced diet) were wrong, and government dietary guidelines need rewriting.If CIM is wrong: you have been sold a simplified causal story and missed the actually useful levers (total energy + protein + food quality + sustainability).
Why the debate matters: not to pick a side, but to see the chasm between a mechanism story and clinical evidence. A mechanism that sounds airtight on paper can completely collapse inside a metabolic ward. Over the next three scenes we walk the evidence chain.
立场 · 两派各自押的是什么
能量平衡模型 (Energy Balance Model, EBM) 是主流共识, 代表人物是 NIH 的 Kevin Hall 和 Dale Schoeller。它并不否认激素在起作用, 只是把激素放在中间变量的位置: 胰岛素、瘦素、胃饥饿素都在影响你吃多少、动多少, 但它们改变不了进来的减去用掉的这笔账。碳水-胰岛素模型 (Carbohydrate-Insulin Model, CIM) 由哈佛的 David Ludwig 和 Cara Ebbeling 提出 (Ludwig 2018)。它把因果链整个倒过来: 不是你吃多了所以胖, 而是你吃的东西让你不得不吃多。高升糖碳水 → 餐后胰岛素飙升 → 脂肪储存激增 → 血糖低谷加饥饿 → 多吃 → 肥胖。推论很干脆: 限碳水、生酮才是减脂的正路, 单纯计算热量是徒劳的。
为什么这场争论值得普通人知道
卖东西的几乎全部押在 CIM 上: 营销话术就那么几句 —— 碳水让你胖、胰岛素是肥胖的总开关、不计卡路里只戒糖、生酮治愈一切 —— 听起来比少吃多动性感得多, 卖书、卖课、卖低碳食品都好做。如果 CIM 是对的: 主流公共健康建议 (减少总热量、平衡饮食) 几十年都错了, 政府膳食指南需要重写。如果 CIM 是错的: 你被卖了一个简化的因果故事, 同时错过了真正有用的杠杆 —— 总能量、蛋白、食物质量、可持续性。
所以这一岛真正要教的不是哪派赢了, 而是机制叙事和临床证据之间的鸿沟: 一个听起来无比合理的机制, 在严格的代谢病房实验里可能完全站不住脚。
Chapter 2
CIM hypothesis · a plausible mechanism story
CIM hypothesis · a plausible mechanism story
The plain version first: the 'carbohydrate-insulin hypothesis' claims that what makes you fat isn't eating too much — it's a loop: refined carbs → insulin spikes → energy gets locked into fat → you're hungry again soon → you eat more. Every step sounds plausible, but it rests on one testable prediction that the next scene overturns. First, the story on its own terms.
Ludwig & Ebbeling 2018 (JAMA Internal Medicine) turned the carbohydrate-insulin hypothesis into a heavily-cited position paper. The mechanism story runs as follows.
Step 1 · High-GI foods spike blood glucose
A bowl of white rice / glass of juice → glucose floods the bloodstream within 30-60 minutesInsulin is secreted in bulk to push glucose into cells
Step 2 · Insulin = fat-storage signal
Insulin suppresses adipose lipolysis (↓ HSL) → already-stored fat cannot exitInsulin activates lipoprotein lipase: An enzyme on the vessel wall that unloads blood fat into muscle or fat tissue. → circulating triglycerides are pulled into adipocytesHepatic de novo lipogenesis (DNL) is stimulated → excess glucose is converted to fatNet effect: energy is forcibly routed into fat storage rather than burned by muscle
Step 3 · Glucose nadir → hunger → overeating
Insulin overshoot → reactive hypoglycemia 1-3 hours laterThe brain reads 'low energy' → fires hunger signals (↑ ghrelin, blunted leptin response)You feel hungry an hour earlier than expected, reach for the next cookie → cycle amplifies
Step 4 · Corollary
You are not fat because you eat too much; you eat too much because of what you eatTherapy: carb restriction / ketogenic → insulin stays low → fat can be released + hunger disappears → you naturally eat less
Why this story spreads so well:
Every step rests on real biochemistry (insulin does suppress lipolysis — no one disputes that piece)Short-term weight-loss data look supportive: low-carb / keto produce fast losses in the first 1-2 weeksBut — the bulk of that early loss is water and glycogen, not fat. Each gram of liver glycogen binds 3-4 g of water. Cut carbs → glycogen depletes → 2-3 kg lighter in a week, none of which is fat.
CIM's core falsifiable prediction is: at equal calories, low-carb should lose more fat than low-fat. Kevin Hall dragged that sentence into a metabolic ward.
Ludwig & Ebbeling 2018 (JAMA Internal Medicine) turned the carbohydrate-insulin hypothesis into a heavily-cited position paper. The mechanism story runs as follows.
Step 1 · High-GI foods spike blood glucose
A bowl of white rice / glass of juice → glucose floods the bloodstream within 30-60 minutesInsulin is secreted in bulk to push glucose into cells
Step 2 · Insulin = fat-storage signal
Insulin suppresses adipose lipolysis (↓ HSL) → already-stored fat cannot exitInsulin activates lipoprotein lipase: An enzyme on the vessel wall that unloads blood fat into muscle or fat tissue. → circulating triglycerides are pulled into adipocytesHepatic de novo lipogenesis (DNL) is stimulated → excess glucose is converted to fatNet effect: energy is forcibly routed into fat storage rather than burned by muscle
Step 3 · Glucose nadir → hunger → overeating
Insulin overshoot → reactive hypoglycemia 1-3 hours laterThe brain reads 'low energy' → fires hunger signals (↑ ghrelin, blunted leptin response)You feel hungry an hour earlier than expected, reach for the next cookie → cycle amplifies
Step 4 · Corollary
You are not fat because you eat too much; you eat too much because of what you eatTherapy: carb restriction / ketogenic → insulin stays low → fat can be released + hunger disappears → you naturally eat less
Why this story spreads so well:
Every step rests on real biochemistry (insulin does suppress lipolysis — no one disputes that piece)Short-term weight-loss data look supportive: low-carb / keto produce fast losses in the first 1-2 weeksBut — the bulk of that early loss is water and glycogen, not fat. Each gram of liver glycogen binds 3-4 g of water. Cut carbs → glycogen depletes → 2-3 kg lighter in a week, none of which is fat.
CIM's core falsifiable prediction is: at equal calories, low-carb should lose more fat than low-fat. Kevin Hall dragged that sentence into a metabolic ward.
机制 · 四步链条的完整版
Ludwig & Ebbeling 2018 (JAMA Internal Medicine) 把碳水-胰岛素假说写成了一篇高引论文。它的四步是这样接起来的。第一步 · 高升糖食物撞击血糖
一碗白米饭、一杯果汁, 葡萄糖在 30-60 分钟内灌入血液。胰岛素为响应而大量分泌, 把葡萄糖压进细胞。
第二步 · 胰岛素就是储脂信号
胰岛素抑制脂肪组织脂解 (把拆解脂肪的那把酶 HSL 压下去) → 已经储存的脂肪出不来胰岛素激活脂蛋白脂肪酶 lipoprotein lipase: An enzyme on the vessel wall that unloads blood fat into muscle or fat tissue. → 血液里的甘油三酯被脂肪细胞抓进去同时刺激肝脏从头合成脂肪 (de novo lipogenesis, DNL) → 多余葡萄糖变脂肪净效应: 能量被强制路由到脂肪储存, 而不是被肌肉燃烧
这三条要连起来看才有分量: 一条堵住出口, 一条打开入口, 一条还在往里加料。三个方向同时指向仓库, 所以 CIM 才敢说能量被锁住了。
第三步 · 血糖低谷 → 饥饿 → 多吃
胰岛素过冲, 1-3 小时后出现反应性低血糖。大脑读到能量不足, 释放饥饿信号 (胃饥饿素 ghrelin 升高, 瘦素 leptin 的饱腹反馈变迟钝)。你提前 1 小时又饿了, 抓下一块饼干 → 循环放大。
第四步 · 推论
不是你吃得多让你胖, 而是你吃的东西让你不得不吃得多解药: 限碳水、生酮 → 胰岛素维持低位 → 脂肪可以释放 + 不会饿 → 自然吃得少
这个故事为什么能流行: 每一步都有真实生化基础 —— 胰岛素确实抑制脂解, 这一段没人反对。真正的争议不在胰岛素会不会做这些事, 而在它做这些事的净结果, 大到足以决定你胖不胖吗。
误区 · 头两周掉的那几公斤, 大半不是脂肪
低碳和生酮在开头掉得快, 这是很多人相信 CIM 的直接体感来源。但掉的东西不对。短期减重数据看似支持: 低碳/生酮在头 1-2 周减重确实快。
但快速减重的主要成分是水和糖原, 不是脂肪。 肝脏把葡萄糖存成糖原时, 每 1 g 肝糖原会拉着 3-4 g 水一起存进去 —— 糖原是亲水的, 它在组织里本来就是泡着的状态。切断碳水, 糖原被优先取用而不再补回, 那些水也跟着一起排掉, 一周轻 2-3 kg。体重秤读数很好看, 但这部分不是脂肪流失。
这解释了两件事: 一是为什么生酮的神奇头两周几乎人人都有; 二是为什么恢复吃碳水后体重会反弹得莫名其妙快 —— 糖原重新装满, 水跟着回来, 而这中间脂肪其实没怎么动过。
所以想判断一种吃法有没有真的减脂, 头两周的体重变化几乎不能用。要么看更长的时间窗, 要么干脆直接量体成分 —— 这正是下一幕那些实验做的事。
CIM 真正可以被证伪的那句话是: 同样的热量, 低碳应该比低脂减更多脂肪。 这句话不看体重秤、不看水, 只看脂肪本身, 所以它可以被实验直接回答。
Chapter 3
Hall counterattack · metabolic-ward RCTs refute CIM
Hall counterattack · metabolic-ward RCTs refute CIM
Kevin Hall (NIH) is the most rigorous voice on the opposing side. Instead of relying on mechanism stories, he puts people into metabolic wards — food precisely weighed, activity controlled, 24-hour indirect calorimetry for energy expenditure, body composition by DXA + labeled water.
Hall 2015 (Cell Metabolism) · isocaloric head-to-head
Design: 17 obese adults, 28-day ward stay, crossover, strictly isocaloricDiet A: carb cut 30%, fat unchanged (low-carb vs baseline)Diet B: fat cut 30%, carb unchanged (low-fat vs baseline)CIM prediction: low-carb should lose more fatActual result: low-fat lost ~ 80 g more fat per day, statistically significantConclusion: under isocaloric conditions, cutting fat — not carbs — produced more measured fat loss via direct fat-balance measurement. CIM's core assumption (low-carb = fat-storage advantage) fails.
Hall 2021 (Nature Medicine) · ad-libitum version
Design: 20 healthy adults, 2-week ward stay per arm, crossoverDiet A: low-fat plant-based (75% carb, 10% fat)Diet B: low-carb animal-based ketogenic (10% carb, 75% fat)Key: ad libitum — eat as much as you wantCIM prediction: keto group, free of insulin spikes, should feel less hunger and eat lessActual result: low-fat plant group ate 689 kcal/day less, lost more weight. The keto group showed no legendary 'ultra-low hunger' — they ate more.
Hall's bonus finding:
Hall 2019 UPF RCT (Cell Metabolism): matched for energy density and macronutrient profile, the ultra-processed-food arm ate 508 kcal/day more than the unprocessed arm, gaining 0.9 kg in 14 days. Food quality and processing dominate intake far more than macronutrient ratios.
Bottom line: not one CIM prediction has been verified in tightly controlled RCTs. That doesn't mean low-carb / keto is ineffective — they do produce weight loss in many people — but the mechanism is not what CIM claims.
Hall 2015 (Cell Metabolism) · isocaloric head-to-head
Design: 17 obese adults, 28-day ward stay, crossover, strictly isocaloricDiet A: carb cut 30%, fat unchanged (low-carb vs baseline)Diet B: fat cut 30%, carb unchanged (low-fat vs baseline)CIM prediction: low-carb should lose more fatActual result: low-fat lost ~ 80 g more fat per day, statistically significantConclusion: under isocaloric conditions, cutting fat — not carbs — produced more measured fat loss via direct fat-balance measurement. CIM's core assumption (low-carb = fat-storage advantage) fails.
Hall 2021 (Nature Medicine) · ad-libitum version
Design: 20 healthy adults, 2-week ward stay per arm, crossoverDiet A: low-fat plant-based (75% carb, 10% fat)Diet B: low-carb animal-based ketogenic (10% carb, 75% fat)Key: ad libitum — eat as much as you wantCIM prediction: keto group, free of insulin spikes, should feel less hunger and eat lessActual result: low-fat plant group ate 689 kcal/day less, lost more weight. The keto group showed no legendary 'ultra-low hunger' — they ate more.
Hall's bonus finding:
Hall 2019 UPF RCT (Cell Metabolism): matched for energy density and macronutrient profile, the ultra-processed-food arm ate 508 kcal/day more than the unprocessed arm, gaining 0.9 kg in 14 days. Food quality and processing dominate intake far more than macronutrient ratios.
Bottom line: not one CIM prediction has been verified in tightly controlled RCTs. That doesn't mean low-carb / keto is ineffective — they do produce weight loss in many people — but the mechanism is not what CIM claims.
数字 · 等热量头对头的那一场
Hall 2015 (Cell Metabolism) · 等热量头对头设计: 17 名肥胖男女, 病房 28 天, 两段交叉, 严格等热量饮食 A: 减少碳水 30%, 脂肪不变 (低碳 vs 基线)饮食 B: 减少脂肪 30%, 碳水不变 (低脂 vs 基线)CIM 预测: 低碳应该多流失脂肪实际结果: 低脂组每日多流失脂肪 ~ 80 g, 而且差异在统计学上显著
交叉设计是这里的关键: 同一个人先后吃两种饮食, 自己当自己的对照。基础代谢率、活动习惯、肠道菌群这些人与人之间差异极大的变量, 于是全部被抵消掉 —— 剩下的差别只可能来自饮食本身。
结论: 等热量条件下, 减脂肪而不是减碳水, 反而直接通过脂肪平衡测量显示出更多的脂肪流失。CIM 的核心假设 (低碳 = 脂肪储存优势) 不成立。
值得注意的是差距的方向, 而不只是大小。CIM 预测的是低碳那一侧更多, 实测是低脂那一侧更多。一个假说的预测被做反了, 比预测幅度不够大要严重得多 —— 后者还能用效应量小解释过去, 前者不能。
数字 · 敞开吃的那一场
Hall 2021 (Nature Medicine) · 自由进食版设计: 20 名健康成人, 病房 2 周一段交叉饮食 A: 低脂植物饮食 (75% 碳水, 10% 脂肪)饮食 B: 低碳动物生酮 (10% 碳水, 75% 脂肪)关键: 自由进食, 想吃多少吃多少CIM 预测: 生酮组因为没有胰岛素飙升, 应该更不饿, 自然吃得少实际结果: 低脂植物组每日少吃 689 kcal, 体重下降更多。生酮组没有传说中的超低饥饿, 反而摄入更高。
这一场比上一场更致命, 因为它测的正是 CIM 最核心的那条行为主张。等热量实验只能回答同样热量下脂肪去了哪里; 自由进食实验回答的是你会自己吃多少 —— 而你会自己吃多少才是 CIM 用来解释肥胖的那一环。把这一环松开让它自由发挥, 它走的方向仍然是反的。
还有一个常被忽略的细节: 生酮组的胰岛素水平确实低得多。也就是说, 机制那一半是对的, 结果那一半不对。胰岛素确实按 CIM 说的动了, 但人并没有因此少吃。这正是机制叙事最容易骗人的地方 —— 中间步骤全对, 不保证终点也对。
数字 · 顺手捡到的那个彩蛋
Hall 2019 UPF RCT (Cell Metabolism): 在能量密度、宏量分配都配平的前提下, 超加工食品组比未加工组每日多吃 508 kcal, 14 天多重 0.9 kg。请注意这一场里没有换宏量比例 —— 碳水、脂肪、蛋白的比例两边一样, 能量密度也一样。唯一变的是食物的形态: 一边软、好嚼、入口即化; 另一边要咀嚼、要拆解、体积更大。结果是进食速度更快、饱腹信号来得更晚, 人在同样的时间里塞进去更多。
所以真正决定你吃多少的杠杆, 排在宏量比例前面的是: 食物长什么样、要嚼多久、密度多高、有多容易一口接一口。这也解释了为什么低碳有效和低脂植物有效可以同时成立 —— 两种吃法在现实里都会把超加工食品挤出去一大半。
这一整幕的总结: CIM 的预测在严格 RCT 里没有一项被验证。低碳仍然可以是一个好选择, 但理由要换掉 —— 换成下一幕说的那个。
Chapter 4
Mainstream consensus · what actually drives weight loss
Mainstream consensus · what actually drives weight loss
After a decade of debate, mainstream consensus has converged (Hall, Schoeller, AHA, ADA, EASD):
① Energy balance is the determinant
Over long horizons (months+), energy deficit is the root cause of fat lossInsulin / hormones are intermediate variables — they influence intake and expenditure but cannot violate energy conservation'I eat nothing and still don't lose' is almost never metabolic dysfunction; it is intake underestimation + expenditure overestimation (Hall 2017 review)
② Macronutrient ratio is a secondary variable
At equal energy deficit, differences in fat loss across macronutrient ratios are small in RCTs (often < 2 kg over 12 months)The real importance of macro ratios lies in: satiety / food preference / sustainability — these influence total intake indirectly
③ DIETFITS Stanford 2018 (JAMA · Gardner et al.) · 12-month head-to-head
Design: 609 overweight adults randomized to healthy low-fat vs healthy low-carb, both emphasizing whole foods + reduced sugar + reduced refined grains + no calorie limitResult: nearly identical 12-month weight loss — low-fat −5.3 kg, low-carb −6.0 kg (not significant)Pre-specified genotype / insulin-secretion responder differences = 0 (CIM's precision-nutrition prediction also failed)Key insight: diet quality matters more than diet type
④ Low-carb works for some people — because it is easier to stick with, not metabolic magic
People with high insulin resistance / diabetes / preference for satiating high-fat-protein meals → better adherence → naturally eat lessEffect real, mechanism mundane: total energy drops, not because they 'bypassed insulin the master switch'
⑤ Low-fat plant works for other people
Those with high cardiovascular risk / who enjoy carbs / already eat plenty of produce → low-fat high-fiber is satiating and benefits cardiovascular health simultaneously
⑥ Protein matters (cross-camp consensus)
0.8-1.0 g/lb body weight (≈ 1.6-2.2 g/kg) preserves lean mass + increases satiety during a deficit (Longland 2016)This holds whether you side with EBM or CIM
Conclusion: nutrition has no 'macro split that works best for everyone'. Energy deficit + food quality + adequate protein + something you can stick with = the actual four-piece toolkit.
① Energy balance is the determinant
Over long horizons (months+), energy deficit is the root cause of fat lossInsulin / hormones are intermediate variables — they influence intake and expenditure but cannot violate energy conservation'I eat nothing and still don't lose' is almost never metabolic dysfunction; it is intake underestimation + expenditure overestimation (Hall 2017 review)
② Macronutrient ratio is a secondary variable
At equal energy deficit, differences in fat loss across macronutrient ratios are small in RCTs (often < 2 kg over 12 months)The real importance of macro ratios lies in: satiety / food preference / sustainability — these influence total intake indirectly
③ DIETFITS Stanford 2018 (JAMA · Gardner et al.) · 12-month head-to-head
Design: 609 overweight adults randomized to healthy low-fat vs healthy low-carb, both emphasizing whole foods + reduced sugar + reduced refined grains + no calorie limitResult: nearly identical 12-month weight loss — low-fat −5.3 kg, low-carb −6.0 kg (not significant)Pre-specified genotype / insulin-secretion responder differences = 0 (CIM's precision-nutrition prediction also failed)Key insight: diet quality matters more than diet type
④ Low-carb works for some people — because it is easier to stick with, not metabolic magic
People with high insulin resistance / diabetes / preference for satiating high-fat-protein meals → better adherence → naturally eat lessEffect real, mechanism mundane: total energy drops, not because they 'bypassed insulin the master switch'
⑤ Low-fat plant works for other people
Those with high cardiovascular risk / who enjoy carbs / already eat plenty of produce → low-fat high-fiber is satiating and benefits cardiovascular health simultaneously
⑥ Protein matters (cross-camp consensus)
0.8-1.0 g/lb body weight (≈ 1.6-2.2 g/kg) preserves lean mass + increases satiety during a deficit (Longland 2016)This holds whether you side with EBM or CIM
Conclusion: nutrition has no 'macro split that works best for everyone'. Energy deficit + food quality + adequate protein + something you can stick with = the actual four-piece toolkit.
数字 · 一整年的头对头, 结果几乎打平
DIETFITS Stanford 2018 (JAMA · Gardner 团队) · 头对头 12 个月设计: 609 名超重成人, 随机分到健康低脂 vs 健康低碳, 两组都强调全食物 + 限糖 + 限精制谷物 + 不限热量结果: 12 个月体重下降几乎相同 — 低脂 −5.3 kg, 低碳 −6.0 kg (差异不显著)预设的基因型、胰岛素分泌型应答差异 = 0 (CIM 的精准营养预测也没站住)
最后那一条值得单独看一眼。CIM 阵营后来的退守阵地是精准营养: 就算平均没差别, 至少胰岛素分泌高的人该更适合低碳吧? DIETFITS 事先就把这个假设写进了方案, 按基因型和胰岛素分泌型分组去找差异 —— 找不到。事先声明再去找和事后从数据里翻出来是两回事; 前者找不到, 分量重得多。
关键洞察: 饮食质量比饮食类型更重要。两组都被要求吃全食物、少糖、少精制谷物, 于是两组都瘦了 —— 这个共同点的贡献, 明显大于碳水还是脂肪这个不同点。
机制 · 宏量比例为什么只能排第二
能量守恒不是一句口号, 它是这场辩论的地板。我什么都不吃也不瘦几乎从来不是代谢异常, 而是摄入低估 + 消耗高估 (Hall 2017 综述)。人对自己吃了多少的估计是系统性偏低的, 对自己动了多少的估计是系统性偏高的; 两个方向叠加, 就足以吃掉一个本来存在的赤字。这也是为什么代谢病房这么贵还非做不可 —— 它买的就是把这两个误差同时关掉。
在能量赤字相同的前提下, 不同宏量比例对脂肪流失的差异在 RCT 里很小 (常 < 2 kg / 12 个月)。差异之所以小, 是因为宏量比例的作用路径是间接的: 它先影响饱腹感和依从性, 再由这两样去影响总摄入。一条要绕两道弯的路径, 效应量自然做不大。
低碳对某些人有效 — 因为它好坚持, 不是代谢魔法
胰岛素抵抗明显、糖尿病、或者本来就偏好高脂高蛋白那种扛饿口感的人, 在低碳上依从性常常更好, 于是自然吃得少。效用真实, 机制平凡: 总能量下降了, 不是因为绕过了胰岛素这个总开关。
低脂植物对另一些人有效
心血管风险高、喜欢碳水、本来蔬果就吃得多的人, 低脂高纤维的组合饱腹感很强, 心血管同时受益。
蛋白要够 (跨派共识)
0.8-1.0 g/lb 体重 (≈ 1.6-2.2 g/kg) 在减重期保持瘦体重 + 增加饱腹感 (Longland 2016)这一条无论你站 EBM 还是 CIM 都一样 —— 因为它起作用的地方不在能量去了哪里, 而在掉下来的那部分是脂肪还是肌肉
Chapter 5
Practical takeaway · don't get hijacked by extremes
Practical takeaway · don't get hijacked by extremes
With the debate walked through, the practical implications are clear:
① No macro split is 'best for everyone'
Anyone claiming 'this diet is optimal' — keto evangelists or low-fat plant advocates alike — is oversimplifyingDIETFITS 12-month head-to-head: mean difference < 1 kg. Picking the 'wrong' camp is not why you aren't losing.
② Pick what you can sustain long-term
Weight-loss drugs / short fasts / 7-day keto / 7-day cleanses can drop 2 kg next weekThe diet that keeps you 5 kg lighter next year = the good diet. Whether you can sustain it depends far more on lifestyle / preference / culture than on macronutrient ratiosDecision frame: ask — 'will I still be eating this way 3 years from now?' If no, switch.
③ Watch for extreme rhetoric
'Keto cures everything' — no disease is cured by a macro split alone. Keto has legitimate indications in pediatric epilepsy and some metabolic conditions, but it is not universal.'Carbs make you fat' — rice / oats / sweet potatoes / whole wheat / fruit are staples in every long-lived population. People get fat from chronic energy surplus + processed food + sedentary living, not from the carbohydrate molecule.'Insulin is the master switch of obesity' — insulin is one hormone that influences metabolic routing, not a storage demon. Protein also stimulates insulin (some proteins more than carbs). Chicken breast does not make you fat.'Calories are a scam' — calorimetry has been validated repeatedly in metabolic-ward RCTs. Daily counts have measurement error (labels ± 20%, self-tracking ± 30%), but the physics is sound.
④ Protein matters — regardless of camp
During deficit: 1.6-2.2 g/kg body weight; maintenance: 1.2-1.6 g/kgPreserves lean mass + boosts satiety + high thermic effect (TEF 20-30% vs carbs 5-10% vs fat 0-3%)
Atlas connections:
weight-management-foundations (full version of the energy-balance four-piece toolkit)protein-during-deficit (concrete protein execution during a deficit)protein + carbs-fiber + dietary-fats (deep dive on each macro node)ultra-processed-foods (the full Hall 2019 UPF RCT story)insulin-resistance (mechanism of insulin resistance — what CIM did get right)
Closing line: No matter how elegant a mechanism story sounds, it has to clear an RCT. Anyone selling you a shortcut around the body's physics is not selling science.
① No macro split is 'best for everyone'
Anyone claiming 'this diet is optimal' — keto evangelists or low-fat plant advocates alike — is oversimplifyingDIETFITS 12-month head-to-head: mean difference < 1 kg. Picking the 'wrong' camp is not why you aren't losing.
② Pick what you can sustain long-term
Weight-loss drugs / short fasts / 7-day keto / 7-day cleanses can drop 2 kg next weekThe diet that keeps you 5 kg lighter next year = the good diet. Whether you can sustain it depends far more on lifestyle / preference / culture than on macronutrient ratiosDecision frame: ask — 'will I still be eating this way 3 years from now?' If no, switch.
③ Watch for extreme rhetoric
'Keto cures everything' — no disease is cured by a macro split alone. Keto has legitimate indications in pediatric epilepsy and some metabolic conditions, but it is not universal.'Carbs make you fat' — rice / oats / sweet potatoes / whole wheat / fruit are staples in every long-lived population. People get fat from chronic energy surplus + processed food + sedentary living, not from the carbohydrate molecule.'Insulin is the master switch of obesity' — insulin is one hormone that influences metabolic routing, not a storage demon. Protein also stimulates insulin (some proteins more than carbs). Chicken breast does not make you fat.'Calories are a scam' — calorimetry has been validated repeatedly in metabolic-ward RCTs. Daily counts have measurement error (labels ± 20%, self-tracking ± 30%), but the physics is sound.
④ Protein matters — regardless of camp
During deficit: 1.6-2.2 g/kg body weight; maintenance: 1.2-1.6 g/kgPreserves lean mass + boosts satiety + high thermic effect (TEF 20-30% vs carbs 5-10% vs fat 0-3%)
Atlas connections:
weight-management-foundations (full version of the energy-balance four-piece toolkit)protein-during-deficit (concrete protein execution during a deficit)protein + carbs-fiber + dietary-fats (deep dive on each macro node)ultra-processed-foods (the full Hall 2019 UPF RCT story)insulin-resistance (mechanism of insulin resistance — what CIM did get right)
Closing line: No matter how elegant a mechanism story sounds, it has to clear an RCT. Anyone selling you a shortcut around the body's physics is not selling science.
误区 · 最响的那四句, 逐条拆
生酮治愈百病 — 没有任何疾病能被一种宏量分配治愈。生酮在 1 型癫痫儿童 + 某些代谢综合征有合法适应症, 但那是适应症, 不是万能。碳水让你胖 — 米饭、燕麦、红薯、全麦、水果在所有长寿地区都是主食。让人胖的是长期能量过剩 + 加工食品 + 久坐, 不是碳水分子。同一个分子装在不同的食物里, 后果可以完全不同: 白砂糖和红薯里的都是碳水, 但一个几乎不用嚼、饱腹感极低, 另一个体积大、有纤维、进食速度被迫慢下来。
胰岛素是肥胖总开关 — 胰岛素是一个影响代谢路由的激素, 不是储脂魔王。这里有个反例很好用: 蛋白质同样刺激胰岛素, 一些蛋白甚至比碳水刺激得更多。如果胰岛素真是那个总开关, 吃鸡胸肉就该让人发胖 —— 显然不会。
卡路里是骗局 — 卡路里测量在病房 RCT 里被反复验证。它确实有日常计量误差 (标签 ± 20%, 自我估算 ± 30%), 但测不准和不存在是两回事。体重秤也有误差, 没人因此说体重是骗局。
这四句有一个共同的形状: 都把一条多因素的链条压缩成一个开关, 而开关是最好卖的东西 —— 它承诺你只要拨一下就完事。真实的身体里没有这种开关。
执行 · 蛋白吃多少, 以及为什么是它
减重期 1.6-2.2 g/kg 体重, 维持期 1.2-1.6 g/kg。这一条无论你站哪一派都成立。蛋白在减重期做三件事:
保住肌肉。 能量赤字里身体会同时拆脂肪和拆肌肉; 蛋白摄入足够时, 拆下来的那部分里肌肉的占比明显下降。掉下来的是脂肪还是肌肉, 决定了你瘦下来之后是什么样子, 也决定了你的基础代谢会不会跟着往下掉。更扛饿。 同样热量下, 蛋白带来的饱腹感明显高于碳水和脂肪, 于是你在不刻意克制的情况下也会自然少吃 —— 又绕回了总能量这条主线。消化它本身就烧掉一部分。 食物热效应 (TEF) 是身体消化吸收这口食物要花掉的能量: 蛋白 20-30% vs 碳水 5-10% vs 脂肪 0-3%。也就是说, 蛋白里有五分之一到接近三分之一在过路时被收了税。
相连的话题:
weight-management-foundations (能量平衡四要素的完整版)protein-during-deficit (减重期蛋白的具体执行)protein + carbs-fiber + dietary-fats (单宏量节点深入)ultra-processed-foods (Hall 2019 UPF RCT 的完整故事)insulin-resistance (胰岛素抵抗的机制 — 那些 CIM 没说错的部分)
References · 6
- Hall, K. D., Bemis, T., Brychta, R., Chen, K. Y., Courville, A., Crayner, E. J., Goodwin, S., Guo, J., Howard, L., Knuth, N. D., Miller, B. V., Prado, C. M., Siervo, M., Skarulis, M. C., Walter, M., Walter, P. J., & Yannai, L. (2015). Calorie for calorie, dietary fat restriction results in more body fat loss than carbohydrate restriction in people with obesity. Cell Metabolism, 22(3), 427–436. 10.1016/j.cmet.2015.07.021
- Hall, K. D., Guo, J., Courville, A. B., Boring, J., Brychta, R., Chen, K. Y., Darcey, V., Forde, C. G., Gharib, A. M., Gallagher, I., Howard, R., Joseph, P. V., Milley, L., Ouwerkerk, R., Raisinger, K., Rozga, I., Schick, A., Stagliano, M., Torres, S., … Chung, S. T. (2021). Effect of a plant-based, low-fat diet versus an animal-based, ketogenic diet on ad libitum energy intake. Nature Medicine, 27(2), 344–353. 10.1038/s41591-020-01209-1
- Gardner, C. D., Trepanowski, J. F., Del Gobbo, L. C., Hauser, M. E., Rigdon, J., Ioannidis, J. P. A., et al. (2018). Effect of low-fat vs low-carbohydrate diet on 12-month weight loss in overweight adults and the association with genotype pattern or insulin secretion: the DIETFITS randomized clinical trial. JAMA, 319(7), 667-679. 609 adults, 12 months; no significant weight-loss difference between healthy low-fat and healthy low-carbohydrate diets, and neither genotype pattern nor baseline insulin secretion predicted which diet worked better. 10.1001/jama.2018.0245
- Hall, K. D., Ayuketah, A., Brychta, R., Cai, H., Cassimatis, T., Chen, K. Y., et al. (2019). Ultra-processed diets cause excess calorie intake and weight gain: an inpatient randomized controlled trial of ad libitum food intake. Cell Metabolism, 30(1), 67-77.e3. 10.1016/j.cmet.2019.05.008
- Longland, T. M., Oikawa, S. Y., Mitchell, C. J., Devries, M. C., & Phillips, S. M. (2016). Higher compared with lower dietary protein during an energy deficit combined with intense exercise promotes greater lean mass gain and fat mass loss: A randomized trial. American Journal of Clinical Nutrition, 103(3), 738–746. 10.3945/ajcn.115.119339
- Hall, K. D. (2017). A review of the carbohydrate-insulin model of obesity. European Journal of Clinical Nutrition, 71(3), 323-326. Several logical consequences of the carbohydrate-insulin model were tested in a pair of carefully controlled inpatient feeding studies and failed; the author concludes that important aspects of the model have been experimentally falsified and the model is too simplistic. 10.1038/ejcn.2016.260