Place · Level 3 · Supplement
Collagen peptides
吃啥补啥最经典误区 · vitamin C 才是真协同 · 关节、皮肤 B 级证据 · 营销 ≫ 证据
Story path
Chapter 1
Triple helix
Triple helix
Collagen is the body's most abundant protein — about 25–30% of total protein. It forms structural scaffolds in skin, bone, cartilage, tendons, ligaments, vessel walls, and sclera.
Its structure is distinctive: a right-handed helix of three α chains, each a repeating (Gly-X-Y)ₙ sequence where:
Glycine (Gly) occupies every third amino acid position — only Gly's tiny side chain fits the center of the helixX position: often proline (Pro)Y position: often hydroxyproline (Hyp) — a post-translational modification adding an -OH to Pro, and this step requires vitamin C as a cofactor
Collagen peptides = partly hydrolyzed collagen — heat plus enzymatic cleavage break the large triple helix into 2–10 kDa short peptides (5–30 amino acids). These short peptides form a gel when cooled (gelatin); further hydrolysis yields cold-water-soluble collagen peptide powder — the form sold by Vital Proteins, Great Lakes, NeoCell, etc.
Practical difference between the two product forms:
Gelatin: partly hydrolyzed; forms a gel in cold water; used in cooking (jellies, mousses) and can be taken orallyCollagen peptide (hydrolyzed collagen): further-hydrolyzed short peptides; dissolves in cold or hot water; no gel feel
The amino acid composition of the two is essentially identical — both are Gly + Pro + Hyp-rich mixtures with 90%+ total protein content. The marketing difference is mainly 'easy-dissolve vs cooking-use', not 'bioactive potency'.
Its structure is distinctive: a right-handed helix of three α chains, each a repeating (Gly-X-Y)ₙ sequence where:
Glycine (Gly) occupies every third amino acid position — only Gly's tiny side chain fits the center of the helixX position: often proline (Pro)Y position: often hydroxyproline (Hyp) — a post-translational modification adding an -OH to Pro, and this step requires vitamin C as a cofactor
Collagen peptides = partly hydrolyzed collagen — heat plus enzymatic cleavage break the large triple helix into 2–10 kDa short peptides (5–30 amino acids). These short peptides form a gel when cooled (gelatin); further hydrolysis yields cold-water-soluble collagen peptide powder — the form sold by Vital Proteins, Great Lakes, NeoCell, etc.
Practical difference between the two product forms:
Gelatin: partly hydrolyzed; forms a gel in cold water; used in cooking (jellies, mousses) and can be taken orallyCollagen peptide (hydrolyzed collagen): further-hydrolyzed short peptides; dissolves in cold or hot water; no gel feel
The amino acid composition of the two is essentially identical — both are Gly + Pro + Hyp-rich mixtures with 90%+ total protein content. The marketing difference is mainly 'easy-dissolve vs cooking-use', not 'bioactive potency'.
Why Gly + Pro + Hyp
Collagen peptides' amino acid composition is the core of their marketing pitch — but the composition itself isn't unique:Glycine (Gly): ~33% of residues. Everywhere in the diet (eggs, meat, soy, gelatin); also synthesized endogenouslyProline (Pro): ~12%. Also not scarceHydroxyproline (Hyp): ~10%. This one is mildly distinctive — dietary 'pre-formed Hyp' only comes from animal collagen, and synthesizing endogenous Hyp requires Pro + vitamin C + 4-prolyl hydroxylase
An interesting fact: your body doesn't take the dietary Hyp you eat and slot it directly into new collagen. The post-translational modification is 'install -OH onto Pro on the spot' — dietary Hyp is broken down in the GI tract into the Pro/Gly pool, and your own fibroblasts later install the modification during translation using vitamin C.
That leads into the next scene's core counterintuitive point: 'eating collagen ≠ directly supplementing skin collagen'.
A worth-remembering sentence: collagen peptides are not 'magic amino acids' — they're just a Gly + Pro-rich partly hydrolyzed protein. If you already hit 1.6–2.2 g/kg/day total protein (consistent with ISSN training-population recommendations), collagen peptides add very little marginal amino acid value.
Chapter 2
'Eat skin to grow skin'
'Eat skin to grow skin'
'Eat collagen → grow collagen' is the supplement world's classic 'eat what you want to grow' fallacy.
Look at the actual process: any protein you eat — beef, eggs, or collagen peptides — goes through this assembly line:
1. Stomach acid + pepsin: shred the triple helix into short peptides
2. Pancreatic trypsin / chymotrypsin / elastase: cut short peptides further into 2–6 amino acid fragments
3. Brush-border peptidases + PEPT1 transporter: absorb di- and tripeptides into enterocytes, where they're further hydrolyzed into free amino acids
4. Portal vein → liver: enter the systemic amino acid pool
The key fact is: the body doesn't preserve 'this was collagen' identity information. All protein is broken down into free amino acids or very short peptides, then reassembled into whatever proteins the cell currently needs.
Eat 20 g of collagen peptides, most of the Gly + Pro disperses into the systemic amino acid pool — and those amino acids can equally well be used to build muscle, enzymes, neurotransmitters, or heme, not just collagen.
So why do some RCTs still see effects? — covered in the next scene and beyond:
1. Specific small peptides (Gly-Pro-Hyp tripeptides etc.) may pass through PEPT1 intact into plasma, where they're recognized at the fibroblast surface as a 'collagen degradation signal mimic' → triggering collagen synthesis upregulation (Shaw 2017 proposed hypothesis, not fully confirmed)
2. Pre-training plus vitamin C: aligns the timing of short peptides in plasma with the tissue synthesis window (exercise-induced collagen turnover)
3. Small samples, surrogate endpoints, clinically meaningful effect mostly absent
So collagen peptides may be slightly better than random protein powder, but they're far from 'directly supplying collagen raw material to the skin'. The marketing narrative tells half the story and frames it as the latter.
Look at the actual process: any protein you eat — beef, eggs, or collagen peptides — goes through this assembly line:
1. Stomach acid + pepsin: shred the triple helix into short peptides
2. Pancreatic trypsin / chymotrypsin / elastase: cut short peptides further into 2–6 amino acid fragments
3. Brush-border peptidases + PEPT1 transporter: absorb di- and tripeptides into enterocytes, where they're further hydrolyzed into free amino acids
4. Portal vein → liver: enter the systemic amino acid pool
The key fact is: the body doesn't preserve 'this was collagen' identity information. All protein is broken down into free amino acids or very short peptides, then reassembled into whatever proteins the cell currently needs.
Eat 20 g of collagen peptides, most of the Gly + Pro disperses into the systemic amino acid pool — and those amino acids can equally well be used to build muscle, enzymes, neurotransmitters, or heme, not just collagen.
So why do some RCTs still see effects? — covered in the next scene and beyond:
1. Specific small peptides (Gly-Pro-Hyp tripeptides etc.) may pass through PEPT1 intact into plasma, where they're recognized at the fibroblast surface as a 'collagen degradation signal mimic' → triggering collagen synthesis upregulation (Shaw 2017 proposed hypothesis, not fully confirmed)
2. Pre-training plus vitamin C: aligns the timing of short peptides in plasma with the tissue synthesis window (exercise-induced collagen turnover)
3. Small samples, surrogate endpoints, clinically meaningful effect mostly absent
So collagen peptides may be slightly better than random protein powder, but they're far from 'directly supplying collagen raw material to the skin'. The marketing narrative tells half the story and frames it as the latter.
The Gly-Pro-Hyp tripeptide hypothesis
Why aren't all RCTs null? A mechanism hypothesis:Specific Gly-Pro-Hyp tripeptides in collagen peptides resist proteolytic cleavage in the gut (because adjacent Pro / Hyp form cis-peptide bonds that proteases struggle to cut)These tripeptides are absorbed intact into enterocytes via PEPT1 (the gut epithelial di-/tripeptide transporter), some escape further intracellular hydrolysis, enter the portal vein, and reach the plasmaThese tripeptides are briefly detectable in plasma (Iwai 2005, Shigemura 2018 PK studies)Hypothesis: they reach the fibroblast surface and are recognized as fragments from 'existing collagen that's been degraded' — the cell receives the signal 'nearby collagen broke, time to synthesize new!' → triggers collagen mRNA expression and translation
This is a plausible hypothesis, not a confirmed mechanism — it could explain why pre-training collagen peptides + vitamin C improve collagen synthesis markers in some studies.
But to be honest: this hypothesis doesn't explain every signal of 'collagen peptides improve skin elasticity', and there's no real-time mechanistic human cellular evidence.
You can treat it as a evolving research direction — one step above 'eat skin to grow skin', but far from the supplement-ad claim of 'directly supplementing skin'.
Chapter 3
Vitamin C is the real cofactor
Vitamin C is the real cofactor
The truly rate-limiting cofactor in collagen synthesis is vitamin C — not the collagen peptide itself.
Mechanism (also covered in the vitamin-c / antioxidant scene; the intersection here):
1. Fibroblasts first translate procollagen — containing ordinary Pro residues
2. Prolyl-4-hydroxylase (P4H) in the endoplasmic reticulum hydroxylates Pro at the 4-position → Hyp
3. Lysyl hydroxylase (LH) hydroxylates Lys → Hyl
4. Both enzymes strictly depend on Fe²⁺ + α-ketoglutarate + ascorbate (vitamin C) as cofactors
5. Without vitamin C → P4H is inactive → collagen Hyp content is insufficient → triple helix stability fails → collagen breaks
Clinical evidence: this is the mechanistic basis of scurvy — Age of Sail sailors lacking vitamin C developed bleeding gums, lost teeth, wounds reopened — all reflecting existing collagen being continuously degraded while new collagen cannot stably form.
Shaw 2017 AJCN key RCT:
n = 8 healthy men, 1 hour pre-jumping trainingThree arms compared: placebo vs 5 g gelatin + 50 mg vitamin C vs 15 g gelatin + 50 mg vitamin CPost-training plasma PINP measured (new collagen synthesis marker)High-dose group plasma PINP doubled vs placeboSmall sample, large signal, mechanism-aligned
Lis 2019 IJSNEM follow-up validation: the vitamin C-fortified collagen (gelatin) synergy effect replicates across more models, but dose / timing window matter.
Practical conclusion: if you genuinely want to try collagen peptides, take them with vitamin C (50–100 mg together), 30–60 minutes pre-training, aligning with the connective tissue repair window. Taking collagen peptides without vitamin C is wasting money.
Reverse inference: if you eat 1 orange + 1 egg + 1 chicken breast, you've already got vitamin C + a complete amino acid profile + some Gly/Pro/Hyp. On the 'collagen synthesis raw material' dimension, this is almost equivalent to collagen peptide + vitamin C capsules — with higher nutrient density and satiety on top.
Mechanism (also covered in the vitamin-c / antioxidant scene; the intersection here):
1. Fibroblasts first translate procollagen — containing ordinary Pro residues
2. Prolyl-4-hydroxylase (P4H) in the endoplasmic reticulum hydroxylates Pro at the 4-position → Hyp
3. Lysyl hydroxylase (LH) hydroxylates Lys → Hyl
4. Both enzymes strictly depend on Fe²⁺ + α-ketoglutarate + ascorbate (vitamin C) as cofactors
5. Without vitamin C → P4H is inactive → collagen Hyp content is insufficient → triple helix stability fails → collagen breaks
Clinical evidence: this is the mechanistic basis of scurvy — Age of Sail sailors lacking vitamin C developed bleeding gums, lost teeth, wounds reopened — all reflecting existing collagen being continuously degraded while new collagen cannot stably form.
Shaw 2017 AJCN key RCT:
n = 8 healthy men, 1 hour pre-jumping trainingThree arms compared: placebo vs 5 g gelatin + 50 mg vitamin C vs 15 g gelatin + 50 mg vitamin CPost-training plasma PINP measured (new collagen synthesis marker)High-dose group plasma PINP doubled vs placeboSmall sample, large signal, mechanism-aligned
Lis 2019 IJSNEM follow-up validation: the vitamin C-fortified collagen (gelatin) synergy effect replicates across more models, but dose / timing window matter.
Practical conclusion: if you genuinely want to try collagen peptides, take them with vitamin C (50–100 mg together), 30–60 minutes pre-training, aligning with the connective tissue repair window. Taking collagen peptides without vitamin C is wasting money.
Reverse inference: if you eat 1 orange + 1 egg + 1 chicken breast, you've already got vitamin C + a complete amino acid profile + some Gly/Pro/Hyp. On the 'collagen synthesis raw material' dimension, this is almost equivalent to collagen peptide + vitamin C capsules — with higher nutrient density and satiety on top.
Baar lab's loading-pulse protocol
Keith Baar's UC Davis muscle physiology lab is the source of the Shaw 2017 / Lis 2019 research line — and the practical protocol he proposed has been widely replicated in athletic circles:Protocol:
1. 30–60 minutes before training: take 5–15 g collagen peptides (or equivalent gelatin) + 50 mg vitamin C — timed so Gly-Pro-Hyp tripeptides + ascorbate peak in plasma when the tissue synthesis window is open
2. Perform short-burst stimulus training: typically jumping, plyometrics, isometrics, or light strength work — providing 'mechanical stimulus' to tendons / ligaments / joint capsules to trigger fibroblast synthesis response
3. 6–8 hours after training: continue with another divided dose (e.g. take again 6 hours later)
4. Continue for 4–12 weeks
Suitable scenarios:
Jumping / sprinting / basketball / volleyball / tennis: sports with high ligament / Achilles / patellar tendon loadCrossFit / powerlifting: joint capsule repeatedly loadedRecovering from ligament / tendon injury: clinical rehab adjunct (but doesn't replace physical therapy)
Less suitable:
Ordinary endurance running / cycling: connective tissue turnover is low, marginal benefit weakStrength training where MPS is the main task: collagen peptides aren't a complete protein (lacking tryptophan) — they're not the MPS (muscle protein synthesis) first choice; whey + leucine is preferred for MPS
One last warning: this is a marginal optimization for people who already have total protein and training dialed in, not a foundation. If you haven't yet sorted out training, sleep, total protein, and vitamin C intake, adding collagen peptides is the lowest-priority lever.
Chapter 4
Joints + skin RCTs
Joints + skin RCTs
Collagen peptides have B-tier clinical evidence in two specific scenarios — neither A-tier nor D-tier. Evaluating supplements has to come down to specific scenarios.
1. Exercise-related joint pain (athletes with joint pain)
Clark 2008 Curr Med Res Opin — the foundational study in this line:
n = 147 college varsity athletes with persistent exercise-related joint pain10 g collagen peptides vs placebo, for 24 weeksResting joint pain reduced ~25% (visual analog scale, P < 0.05)Walking / standing joint pain reduced ~20%No systemic side effectsLarge sample (relative to supplement studies), long duration, clear endpoints — which is why this study has been heavily cited since 2008
Replication studies:
Zdzieblik 2017 and several other small RCTs replicated similar signals in knee / hip pain athlete cohortsKhatri 2021 Amino Acids systematic review: pooled effect exists but small; for hard endpoints in clinical arthritis (OA) (X-ray cartilage thickness, WOMAC scores), the effect is inconsistent
Conclusion: the athlete joint pain signal is real; the clinical osteoarthritis signal is weak; 'preventing joint problems' in healthy people has no data.
2. Skin elasticity / wrinkles
Proksch 2014 Skin Pharmacol Physiol — classic RCT:
n = 69 women, age 35–552.5–5 g specific collagen peptide (Verisol) vs placebo, for 8 weeksSkin elasticity (cutometer) +7–13% vs placeboWrinkle depth, skin hydration, and other secondary endpoints showed weak signals
Replication: multiple small RCTs show consistent signal for skin elasticity; small signal for wrinkles / firmness; almost no signal for anti-photo-aging (UV damage).
Khatri 2021 systematic review: skin effects are B-tier, mainly small-to-moderate improvements in elasticity and hydration — not an anti-wrinkle miracle.
Overall assessment: collagen peptides have B-tier evidence for these two specific endpoints — joint pain plus skin elasticity — more solid than vitamin E / NMN / most 'influencer supplements', but far below the A-tier evidence for creatine / caffeine / vitamin D / fish oil.
Will it make you look 5 years younger? No. Will your skin be noticeably better in 10 years? Evidence doesn't support that. Will a tendinopathy-prone athlete have 20% less training pain for 8–24 weeks? Possibly.
1. Exercise-related joint pain (athletes with joint pain)
Clark 2008 Curr Med Res Opin — the foundational study in this line:
n = 147 college varsity athletes with persistent exercise-related joint pain10 g collagen peptides vs placebo, for 24 weeksResting joint pain reduced ~25% (visual analog scale, P < 0.05)Walking / standing joint pain reduced ~20%No systemic side effectsLarge sample (relative to supplement studies), long duration, clear endpoints — which is why this study has been heavily cited since 2008
Replication studies:
Zdzieblik 2017 and several other small RCTs replicated similar signals in knee / hip pain athlete cohortsKhatri 2021 Amino Acids systematic review: pooled effect exists but small; for hard endpoints in clinical arthritis (OA) (X-ray cartilage thickness, WOMAC scores), the effect is inconsistent
Conclusion: the athlete joint pain signal is real; the clinical osteoarthritis signal is weak; 'preventing joint problems' in healthy people has no data.
2. Skin elasticity / wrinkles
Proksch 2014 Skin Pharmacol Physiol — classic RCT:
n = 69 women, age 35–552.5–5 g specific collagen peptide (Verisol) vs placebo, for 8 weeksSkin elasticity (cutometer) +7–13% vs placeboWrinkle depth, skin hydration, and other secondary endpoints showed weak signals
Replication: multiple small RCTs show consistent signal for skin elasticity; small signal for wrinkles / firmness; almost no signal for anti-photo-aging (UV damage).
Khatri 2021 systematic review: skin effects are B-tier, mainly small-to-moderate improvements in elasticity and hydration — not an anti-wrinkle miracle.
Overall assessment: collagen peptides have B-tier evidence for these two specific endpoints — joint pain plus skin elasticity — more solid than vitamin E / NMN / most 'influencer supplements', but far below the A-tier evidence for creatine / caffeine / vitamin D / fish oil.
Will it make you look 5 years younger? No. Will your skin be noticeably better in 10 years? Evidence doesn't support that. Will a tendinopathy-prone athlete have 20% less training pain for 8–24 weeks? Possibly.
Marketing vs. evidence gap
Collagen peptides have one of the largest marketing-vs-evidence gaps of any supplement category:| Marketing claim | Evidence tier |
|---|---|
| Anti-wrinkle / look 10 years younger | D (no strong evidence) |
| Anti-photo-aging / sun damage protection | D |
| Repairs / rebuilds joint cartilage (clinical OA) | C (inconsistent signal) |
| Hair / nails thicker / longer | C (anecdotes + small studies) |
| Reduces exercise joint pain | B (Clark 2008) |
| Improves skin elasticity (women 35–55) | B (Proksch 2014) |
| Specific small peptides have magical bioactivity (GHK etc.) | D (lab-level, not human-confirmed) |
Vital Proteins (now owned by Nestlé), Garden of Life, NeoCell, Sports Research, Ancient Nutrition collectively generate $2B+ in annual revenue; the global collagen peptide market was valued at $9B+ in 2024 and projected to double by 2032. This is one of the supplement industry's fastest-growing categories, driven mainly by Instagram, KOLs, and 'celebrities are drinking it' — leaving clinical evidence far behind.
Practical: if the claim you hear lands in the D column, close your wallet. If your goal falls in one of the two B-column scenarios, you can try a course, paired with the Baar protocol + vitamin C, observing over 8–24 weeks.
Chapter 5
Decision tree
Decision tree
Do you need collagen peptides?
Q1: what's your goal?
Anti-wrinkle / look 5 years younger: no strong evidence, not a priority — spending on sunscreen + quitting smoking + more produce gives 10× the ROIRepair arthritis / rebuild cartilage: signal is weak — physical therapy + weight management + (when necessary) intra-articular injection beats supplementsAthlete or active person with chronic joint pain / tendon discomfort: B-tier evidence exists, try the Baar protocol for 8–24 weeks35–55-year-old skin elasticity declining, want modest improvement: B-tier evidence exists, try the Proksch doseTotal protein intake insufficient (< 1.2 g/kg): fix total protein first with whey / protein powder / food — collagen peptides aren't the priorityRecovering from ligament / tendon injury: B-tier useful as physical therapy adjunct
Q2: which product form?
Collagen peptide powder (hydrolyzed collagen): dissolves in cold or hot water, no gel, easy to add to coffee / oats; the form most RCTs useGelatin: Baar protocol has used it; cooking-friendly, but gels in cold water, inconvenient for daily useLiquid collagen peptide: marketing premium is high, essentially powder + water + flavoringCollagen capsules: dose too low (a few hundred mg per capsule, you need 5–15 g), extremely cost-inefficient'Specific peptide' patented brands (Verisol / Peptan / TendoForte etc.): the specific peptides Proksch / Clark used. Effect variation across different brands hasn't been adequately validated, but a conservative approach is to choose brands with direct RCT validation
Q3: how to take it?
Baar athlete protocol: 5–15 g + 50 mg vitamin C, 30–60 minutes pre-training, 4–12 weeksProksch skin protocol: 2.5–5 g/day for 8 weeks, then assess skin elasticityClark joint protocol: 10 g/day for 24 weeks, then assess joint painKey: don't stack 'collagen + hyaluronic acid + chondroitin + glucosamine + quercetin' combination powders — doses become unclear, cost-effectiveness drops, interactions unclear
Q4: what not to combine with:
As sole plant protein source: collagen peptides lack tryptophan, can't be your only protein sourcePregnancy / lactation: generally considered safe (animal-source protein), but lacking dedicated RCTsGelatin allergy (rare): mainly pork-source / beef-source protein allergy — switch to fish-sourceChronic kidney disease (CKD): high protein load, use cautiously with physician guidance
Q1: what's your goal?
Anti-wrinkle / look 5 years younger: no strong evidence, not a priority — spending on sunscreen + quitting smoking + more produce gives 10× the ROIRepair arthritis / rebuild cartilage: signal is weak — physical therapy + weight management + (when necessary) intra-articular injection beats supplementsAthlete or active person with chronic joint pain / tendon discomfort: B-tier evidence exists, try the Baar protocol for 8–24 weeks35–55-year-old skin elasticity declining, want modest improvement: B-tier evidence exists, try the Proksch doseTotal protein intake insufficient (< 1.2 g/kg): fix total protein first with whey / protein powder / food — collagen peptides aren't the priorityRecovering from ligament / tendon injury: B-tier useful as physical therapy adjunct
Q2: which product form?
Collagen peptide powder (hydrolyzed collagen): dissolves in cold or hot water, no gel, easy to add to coffee / oats; the form most RCTs useGelatin: Baar protocol has used it; cooking-friendly, but gels in cold water, inconvenient for daily useLiquid collagen peptide: marketing premium is high, essentially powder + water + flavoringCollagen capsules: dose too low (a few hundred mg per capsule, you need 5–15 g), extremely cost-inefficient'Specific peptide' patented brands (Verisol / Peptan / TendoForte etc.): the specific peptides Proksch / Clark used. Effect variation across different brands hasn't been adequately validated, but a conservative approach is to choose brands with direct RCT validation
Q3: how to take it?
Baar athlete protocol: 5–15 g + 50 mg vitamin C, 30–60 minutes pre-training, 4–12 weeksProksch skin protocol: 2.5–5 g/day for 8 weeks, then assess skin elasticityClark joint protocol: 10 g/day for 24 weeks, then assess joint painKey: don't stack 'collagen + hyaluronic acid + chondroitin + glucosamine + quercetin' combination powders — doses become unclear, cost-effectiveness drops, interactions unclear
Q4: what not to combine with:
As sole plant protein source: collagen peptides lack tryptophan, can't be your only protein sourcePregnancy / lactation: generally considered safe (animal-source protein), but lacking dedicated RCTsGelatin allergy (rare): mainly pork-source / beef-source protein allergy — switch to fish-sourceChronic kidney disease (CKD): high protein load, use cautiously with physician guidance
Bone broth question
Classic bone broth marketing narratives:'Ancestral superfood''Natural collagen + minerals''Repairs the gut + anti-inflammatory'
Chemical facts:
Long-simmered bone broth does contain real gelatin (i.e. hydrolyzed collagen) — released through heat and prolonged breakdownBut a single cup (240 ml) of bone broth contains very limited gelatin — typically 1–6 g, depending on bone ratio, simmering time, temperatureTo hit the Baar protocol's 5–15 g, you'd need 1–3 cups — high in calories (fat + sodium)Commercial packaged bone broth is even less transparent
Minerals?
Calcium / magnesium / phosphorus exist in bones, but acid (vinegar etc.) extraction efficiency is limited — a single cup usually has < 10% DV100 g of broccoli provides more minerals than a cup of bone broth
Anti-inflammatory? Gut repair?
These claims are almost entirely indirect extrapolation + influencer amplificationNo hard RCT supports 'drinking bone broth improves leaky gut / SIBO / autoimmune'
Conclusion: bone broth is soup — warm, tasty, with some protein and amino acids, socially meaningful. But it's not a supplement-grade collagen source, not a superfood. If you want 5–15 g of collagen, weighing collagen peptide powder is 10× more efficient and cheaper.
If you enjoy bone broth's taste and ritual, drinking it is fine — just don't treat 'I drink a cup of bone broth daily' as evidence of 'I'm scientifically supplementing collagen'.