Collagen, broth and connective tissue
In this investigation
The question behind the bottle
I prefer drinking broth to taking collagen powder. I have also reported increasing training and front-of-shoulder discomfort that became progressively more uncomfortable over about a week. Those are separate facts: a food preference does not identify the painful tissue or establish that collagen, glycine or a vitamin is missing.
Nutrition can support recovery, but a progressively worsening shoulder deserves assessment and adjustment of the movements that aggravate it. Sudden severe pain, inability to move the arm, marked swelling, persistent numbness or pain after a substantial injury need prompt medical attention. The food log cannot distinguish tendon pain from other shoulder problems. NHS shoulder guidance.
What the famous gelatin experiment showed
In a crossover experiment, eight healthy men received placebo, 5 g or 15 g gelatin with 48 mg vitamin C one hour before six minutes of rope-skipping, repeated three times daily for three days. The 15 g condition roughly doubled the circulating PINP area under the curve. Serum from four participants also increased collagen in engineered ligament tissue. Shaw et al., 2017.
PINP is largely a bone-formation marker, not a direct measure of shoulder-tendon healing. The engineered tissue was a laboratory model. No injury rate, return-to-sport or shoulder-pain benefit was tested. Every drink contained vitamin C, so its separate effect was not isolated. Its registry was entered after 2015 collection and labels registration retrospective.
The evidence did not stop there
Different measurements give different answers. In 45 young recreational athletes, 30 g collagen after resistance exercise increased circulating glycine and proline, but did not significantly increase muscle connective protein synthesis over the five-hour recovery period compared with placebo. Whey significantly increased myofibrillar synthesis relative to placebo; collagen did not, but the direct whey–collagen comparison was also nonsignificant. These were muscle biopsies, a more direct tissue measurement than a circulating marker, but still not an injured tendon or a long-term healing outcome. Aussieker et al., 2023.
A 2025 trial in another 45 young adults compared 30 g hydrolyzed collagen, a matching mixture of its free amino acids and placebo after unilateral resistance exercise. Neither supplement significantly increased muscle connective or myofibrillar synthesis in rested or exercised muscle over six hours. More amino acids in the blood therefore did not automatically mean more synthesis in that tissue. Aussieker et al., 2025.
Longer training trials provide some positive signals. In 50 healthy men, 5 g of specific collagen peptides daily during 14 weeks of resistance training produced greater patellar-tendon cross-sectional area at some measured locations. Tendon stiffness, muscle size and knee-extension strength improved with training without significant between-group differences. In 20 middle-aged men, a different 12-week protocol using 30 g collagen plus 50 mg vitamin C after training improved tendon size and mechanical properties more than placebo plus vitamin C; muscle strength improvements did not significantly differ between groups. These findings concern tendon adaptation, not proof of fewer ruptures or faster recovery from an existing shoulder injury. Jerger et al., 2023, Nulty et al., 2025.
There is also a relevant longer-term null result. In 39 healthy young men, 15 g collagen daily during 15 weeks of lower-body resistance training did not improve tendon size or mechanical adaptation more than placebo. Tendon stiffness rose in both groups, about 17% with collagen and 21% with placebo, without a significant treatment difference. Collagen therefore has not consistently added benefit even for these structural outcomes. Balshaw et al., 2023.
A June 2026 report in 29 young men adds a longer-term muscle-tissue finding. After 12 weeks of resistance training, 15 g daily of specific collagen peptides produced a greater increase in the type I collagen staining signal: 29.8% versus 9.9%. Type III and IV changes did not significantly differ between groups. This measured immunofluorescence intensity in muscle sections, rather than synthesis rate or injury recovery. Its placebo contained no protein or calories, so it cannot establish whether collagen outperforms another added protein. The manufacturer partly funded the study, and one author disclosed collagen-related patents. It is a useful tissue finding alongside the acute synthesis nulls above; the methods and timescales differ. Jerger et al., 2026.
The biopsy paper is a later report from an existing trial cohort, rather than independent replication of its earlier molecular report. Its registry labels registration retrospective and lists long-term histology as secondary; the primary endpoint concerns mTOR activation. This limits claims about prospective confirmation of the selected stain finding. Original trial registry.
Pain and function are a harder test
A manufacturer-funded crossover pilot enrolled 20 people with chronic mid-portion Achilles tendinopathy. Both groups followed a strengthening and return-to-running program. Over the first three months, VISA-A improved 12.6 points with 2.5 g specific collagen peptides twice daily, versus 5.3 with placebo, before treatments crossed over. Praet et al., 2019.
Two participants later withdrew. The current registry lists ultrasound structure as primary and VISA-A as secondary, and labels registration retrospective. This clinical signal needs larger confirmation: crossing treatments without washout during improvement complicates interpretation. A specific Achilles product does not establish broth efficacy or shoulder treatment.
A small randomized study after surgery for large rotator-cuff tears tested collagen, arginine, MSM and bromelain together. Six-month pain ratings and selected MRI grades favored supplementation; one-year functional scores, their pain component and re-tear comparisons did not establish benefit. Patients and surgeons knew allocation. Three participants stopped because of diarrhea, and replacement/flow counts are inconsistent. The combination cannot isolate collagen; a repaired tear differs from my unassessed discomfort. Shoulder-bursa collagen injections are another intervention, not evidence for broth. Gumina et al., 2012, Godek et al., 2022.
Why broth is not a known collagen dose
An analysis of commercial, homemade and café bone broths found substantial differences in collagen-related amino acids. Standardized broths generally supplied less of the relevant amino acids than the 20 g collagen supplement comparator. Some café preparations were richer, but the variability makes “one bottle” an unreliable way to reproduce an experiment. This was a composition study, not evidence that broth improved injuries. Alcock et al., 2019.
A protein label reports total protein, not grams of collagen, glycine or a studied peptide preparation. Without a verified assay, I should log those specific amounts as unknown. Broth can still be an enjoyable food. Compare the amount I will actually drink, protein and sodium per entire package, ingredients and price. A higher price or “bone” on the label does not establish a therapeutic dose, and drinking more solely to reach an assumed dose can add considerable sodium or displace meals.
Check the bottle’s storage label. The meal/storage guide covers opened packages and all-day drinks.
Glycine, vitamin C and the rest of the diet
Glycine is abundant in collagen, but isolated glycine is not proven shoulder-injury treatment. A frequently cited Achilles experiment used a 5% glycine diet in rats. That establishes an animal finding, not a human dose or outcome. A small human sleep-restriction crossover study analyzed only seven participants taking 3 g glycine before bed; it found some next-day fatigue and vigilance improvements, not tendon healing. Vieira et al., 2015, Bannai et al., 2012.
The latest food estimate already contains substantial vitamin C from the pepper and fruit. The gelatin study used tens of milligrams, so its design does not justify adding gram-dose vitamin C to this diet. A collagen product also has a different amino-acid composition from dairy, fish or chicken; its muscle-building value needs its own evidence rather than being inferred from total protein grams.
The practical choice is to keep broth optional because I enjoy it, verify its package rather than assigning a collagen dose, and build recovery around enough food, ordinary protein, sleep and appropriate loading. A collagen experiment would be an adjunct with a defined goal and outcome; it would not make the shoulder diagnosis unnecessary. The study doses here describe research protocols, not a personalized prescription.