Which supplements are worth taking when you’re hypermobile

• About
• Latest Posts

Adam Foster

Director at The Fibro Guy

Adam Foster is the founder of The Fibro Guy and one of the leading voices in hypermobility, EDS, and chronic pain rehabilitation. A former Reconnaissance Soldier, he developed chronic pain after an IED blast in Afghanistan and built his approach from the ground up when the medical system had no answers. Since 2009 he has worked with thousands of people across eight countries, drawing on pain neuroscience, motor learning, and sensorimotor rehabilitation. He co-authored "No Pain, No Pain" with trauma psychologist Dr Mairi Harper, directed the documentary "Invisible Monster," and has been featured in Forbes, BBC Radio 4, ITV News, LiveScience, and The Telegraph.

Latest posts by Adam Foster (see all)

• Which supplements are worth taking when you’re hypermobile - 4 October 2026
• Hypermobile Wrists: Why They Hurt, Which Brace to Use, and How to Build Control - 4 October 2026
• Endometriosis, hypermobility and EDS: where the connective tissue link holds up, and where it doesn’t - 1 October 2026

Nothing you swallow will change the instructions your body builds connective tissue from. That’s the short version, and on its own it’s useless to you, because it leaves you standing in front of a drawer full of supplements with no idea which ones to keep.

So here’s the much longer version. Which of the things you’ve been using has something behind it, which is being sold to you on an argument that merely sounds right, and how to tell the two apart yourself, the next time somebody points you at a supplement, despite how well meaning their intentions are.

The food side of all this lives in our diet article, and in our POTS and salt piece, and creatine has a post of its own.

It’s made of collagen, so take collagen

Most of what gets marketed to our community rests on that phrase.

Name what the tissue is made of, then sell you that substance. Ligaments are largely collagen, so take collagen. Cartilage contains glucosamine and chondroitin, so take those. Collagen needs vitamin C in order to be built properly, so take vitamin C. Every one of those sentences is true in its first half, and none of them establishes the second half.

It borrows the shape of a real argument. The body needs a thing, this product contains the thing, therefore more of the thing is good. The step that gets skipped is whether you were ever short of it, and whether your body has any way of taking a substance that arrives in your stomach and installing it where you want it.

That second step is where nearly all of this falls down. Your digestive system doesn’t do deliveries. It doesn’t open a parcel marked ligament and forward it to your ankle. It takes everything that arrives, breaks it down into its component parts, absorbs those parts into the blood, and then your own cells build whatever your own genes tell them to build, in whatever tissue they were always going to build it in. By the time the raw material gets anywhere near your tissue, it’s anonymous.

So, the question isn’t whether your body uses a substance, it nearly always does. The question is whether you were short of the raw material, and whether a shortage of raw material is the reason the thing isn’t working. Keep that one in the foreground please.

Collagen

Collagen is the one everybody asks about, so let’s tackle it first.

Collagen is a protein. Like every protein you eat it gets taken apart in the gut, first into short chains and then into individual amino acids, and those amino acids are absorbed into the blood as a pool of building blocks. At that point nothing about them says collagen. They’re the same glycine and proline you’d have got off a chicken thigh. Your cells then draw from that pool and assemble proteins according to the instructions in your own genes.

Which means, what comes out, is decided by the instructions and not by the supply. In the inherited connective tissue disorders, the fault sits in the instructions. Sending more raw material to a factory working from a flawed blueprint doesn’t correct the blueprint. That’s why “it’s made of collagen so take collagen” can’t do what the packaging implies. Collagen supplements aren’t a fraud. The thing people buy them hoping for, tighter ligaments and joints that stop giving way, isn’t something a bag of amino acids can deliver.

The authors of ^[1] state that there’s no established disease specific medical therapy for the Ehlers-Danlos syndromes. That’s a reasonable place to start from when somebody tells you otherwise with a discount code attached.

Where the collagen research was done, and on whom

None of that means collagen peptides do nothing. They do several measurable things. The trick is noticing what was measured and in whom, because the people these results came out of are athletes, lifters and osteoarthritic knees, and not anybody whose tissue was assembled from different instructions.

Hydrolysed collagen, which is what’s in nearly every tub on the shelf, is collagen that’s already been chopped into short fragments before it reaches you. Pooling 26 randomised trials, hydrolysed collagen improved skin hydration and elasticity against a dummy pill, with the authors flagging risk of bias and variation between trials as ongoing concerns ^[2]. A real result, and a result about how well skin holds water, which isn’t the thing keeping you awake.

Tendon is more interesting and a good deal messier. One word first, because it decides how you read the next three sentences: stiffness in a tendon isn’t what you feel in your knees at six in the morning, it’s how much the tendon resists being stretched when you load it, and in a tendon that is the direction you want. Combined with resistance training, collagen peptides increased the cross sectional area of the Achilles tendon, meaning it got physically thicker, while its stiffness didn’t shift against a dummy pill ^[3]. In middle aged men doing twelve weeks of resistance training, patellar tendon adaptations were enhanced ^[4]. Pooled across long term training trials, there was no overall effect on tendon stiffness ^[5]. That cluster is a genuine signal rather than a clean one: the tendon gets bigger, and whether it gets stiffer depends on which trial you happen to be reading.

Underneath that sits a mechanistic layer, and it’s where the dose printed on the back of the tub comes from. Gelatin enriched with vitamin C, taken before intermittent activity, raised the circulating markers of collagen synthesis ^[6]. The mechanistic biomarker work suggests a higher dose of hydrolysed collagen can acutely raise those same circulating markers ^[7]. A marker of synthesis floating about in the blood isn’t a stronger ligament, and it’s the kind of outcome that gets converted into a marketing claim somewhere between the paper and the tub.

On pain, which is what most people actually want, the picture is better than the mechanics. Joint pain improved across multiple randomised trials, in athletes, in healthy adults with joint discomfort, and in knee osteoarthritis ^[8][9].

And then there’s the ankle. In athletes with chronic ankle instability, how stable the ankle felt improved and re-injuries went down ^[10]. Measured with an arthrometer, a device that pulls on a joint and records how far it travels, the joint itself was unchanged ^[10]. So the joint wasn’t measurably less loose, and it gave way less.

Both of those being true at once isn’t a contradiction. How much slack is in a joint and how often that joint lets you down are two separate measurements, because whether a joint gives way depends on what your nervous system is doing with it as well as on how much travel there is in the tissue. Which is why an arthrometer reading is a thin way of asking whether somebody’s ankle is working, and why the work we do goes at the signal and the strategy before it goes anywhere near the tissue.

One more, as a lesson in reading a label. Better organisation of the collagen fibres did turn up on a skin biopsy in one randomised trial, and because the product also contained glucosamine, chondroitin, vitamins, minerals and other compounds, it can’t tell you the collagen peptides were what did it ^[11]. When a result comes off a multi ingredient formula, the result belongs to the formula and not to whichever ingredient is printed on the front of the box. That single move accounts for an enormous amount of what you’ll read in a product description.

Now the bit that decides what to do with all of it. Those trials were run in athletes, in middle aged men lifting weights, in resistance trained young men, in people with knee osteoarthritis, and in older adults having their skin measured. None of those groups is built on connective tissue assembled from different instructions, and a number measured in an osteoarthritic knee doesn’t travel to a hypermobile shoulder just because both turn up in the same article. Adults with HSD or EDS have been put through a small single arm feasibility programme, meaning everybody got the same thing and there was nothing to compare it against, using an anti-inflammatory Mediterranean diet with self-management support, and it didn’t show a reduction in pain across the group ^[12]. That programme enrolled 13 people and 8 of them completed it ^[12]. Hold that one very loosely.

So, the steer. If you’re already doing loaded rehab and you fancy trying collagen peptides alongside it, the training plus loading setting is where that tendon signal sits, and it’s a signal about tendon size rather than about joint stability. Take it as a possible small addition to work you’re already doing. If you’re taking it instead of the rehab, or in the hope that your shoulder will stop popping out, that’s money going nowhere.

Normal isn’t the same as enough

A laboratory reference interval is a description of a crowd. Somebody went and measured a great many people, drew a line near the bottom of the spread, and that line became normal. It answers one question, which is whether or not your result is unusual, compared to other people’s. It was never really built to answer whether your result is low enough to be making you feel like you do, and those are genuinely different questions, with genuinely different answers.

And a treatment threshold is a different kind of number altogether. It’s the point at which something in the body measurably changes: where red blood cell production starts to suffer, where the gut starts cranking up absorption because of a shortfall, and where correcting the level actually changes how somebody feels. That number comes from watching what the body does, rather than from comparing you with a crowd.

The two aren’t the same, they aren’t even particularly close, and the one printed on your results letter is nearly always the first kind. When you’re told your bloods are normal, what you’ve really been told is that you aren’t an outlier.

Nobody explains that, which is how people end up walking out of an appointment overruled by a number that was never built to answer the question they asked in the first instance.

Ferritin, and the two numbers nobody tells you apart

Iron is the best worked out example, which is why it’s here early on, though the principle is a good deal wider than iron.

Ferritin is a storage protein, and measuring it in the blood gives you a reasonable proxy for how much iron is in the bank. For a long time the working cutoff for deficiency had been around 15 µg/L ^[13]. That number is specific for the late stage, when the stores are genuinely scraped out, and it misses the earlier part of the slide ^[13].

The physiologically derived numbers sit higher. In healthy non pregnant women, iron deficient red cell production starts appearing somewhere in the mid twenties ^[13]. Approached from the absorption end instead, by looking for the point where the body starts pulling more iron out of food than it used to, that adaptation looked to begin below a ferritin of around 50 µg/L ^[14]. So there’s a stretch above the cutoff on the form where the letter still says normal, though the body has already started compensating ^[13][14].

Then there’s the complication many of you will have worked out on your own. Ferritin is an acute phase reactant, which means it climbs when there’s inflammation in the system, for reasons that have nothing whatever to do with your iron stores. So where a chronic inflammatory process is running, ferritin can read reassuringly normal while the iron actually available for use isn’t. The way around it, is to stop reading ferritin on its own. Guideline oriented reviews in chronic inflammatory disease converge on ferritin under 100 µg/L, or a transferrin saturation under 20%, as the more appropriate screen ^[15][16]. Transferrin is the protein that ferries iron around the blood, and transferrin saturation is simply how much of that carrying capacity currently has iron sitting on it. It earns its place here because it doesn’t climb with inflammation the way ferritin does.

And symptoms don’t wait for a threshold as you will already know. In adolescents with bleeding disorders, fatigue turned up at ferritin levels above 15 µg/L ^[17]. That’s a specific group and not you, so don’t carry the number over. What it tells you though, is that the relationship between a storage number and how wrung out somebody feels is loose enough that a result above the line settles nothing on its own.

So, practically. Don’t go hunting for your own threshold in that list, because those figures came from different groups of people answering different questions and none of them is a line drawn for you. What the list is for is the conversation. Ask what the actual figure was rather than accepting normal, because normal is not a figure. Ask whether transferrin saturation was run alongside it. Ask whether anything inflammatory was going on the week of the test, because if there was, the ferritin may be flattered. And if your figure is sitting in that stretch, between the cutoff on the form and the numbers above it, that’s the point to ask what would be done if it were lower, rather than accepting that nothing will be done because it isn’t.

Every threshold up there was worked out in healthy women, in blood donors, in national survey data and in chronic inflammatory disease cohorts. The authors of ^[18] say that people with hypermobility may be at risk of macronutrient and micronutrient deficiencies, and that the evidence base behind that is rather limited. The gut symptom burden is there to make it plausible: in a large cohort of young people with joint hypermobility, constipation was recorded in 61%, difficulty swallowing in 32%, and dyspepsia or gastroparesis, indigestion and a stomach that empties too slowly, in 25% ^[19]. Eating less, absorbing less and feeling too rough to cook is a reasonable route into a shortfall.

Why nothing budges when you do take it

This is the question underneath all the others, and for iron there’s an actual answer, so we’ll do iron and then widen it out.

Your gut doesn’t simply absorb whatever iron turns up in it. The whole thing is governed by a hormone called hepcidin, which works as a gate. When hepcidin is high the gate sits more closed and less iron crosses from the gut into the blood. Hepcidin goes up in response to inflammation, and it also goes up in response to iron itself, which means a large dose of iron does something slightly strange: it pulls the gate partly shut behind itself, so the next dose, taken a few hours later or the following morning, lands on a less receptive gut than the first one did.

That isn’t a theory about why your ferritin hasn’t moved in two years, it’s the mechanism the dosing research is built upon. Given the same total dose, on alternate days, rather than on consecutive days, iron depleted women in randomised trials had lower hepcidin exposure, and fewer gastrointestinal side effects, without the stores refilling any more slowly ^[20][21]. If you’ve been taking iron every single day for years with nothing to show for it, the schedule is worth raising with whoever prescribed it. Taking less of it more cleverly is a real option rather than a folk remedy.

Then there’s form, which you already discuss among yourselves in more detail than most clinicians do, and you’re right to. Ferrous sulfate is the standard first line because it’s cheap and it works, and constipation, nausea and diarrhoea are common enough with it to wreck adherence ^[22]. An effective supplement nobody can keep taking isn’t an effective supplement. Ferrous bisglycinate (amazing band name by the way) came out with fewer gastrointestinal adverse events when the randomised trials were pooled ^[23]. And the summary position across the oral iron literature is that no single oral iron is clearly superior in every setting ^[24], which is unsatisfying and is also where it sits. So the choice is a tolerability question as much as an efficacy one, and the one that makes you throw up is the wrong one for you whatever it says on paper.

This goes well past iron as well. When something you’re taking isn’t doing anything, it can be failing in four separate places, and they take four different fixes. You might not have been short of it in the first place. It might not be getting absorbed, because of the form, the schedule, or what you’re swallowing it alongside. The dose might be wrong. Or the thing you want it to fix might have nothing to do with a shortage of that substance in the first place.

Working out which one you’re in is mostly a matter of order though. Was it ever measured, and what was the figure. If it was low and you’ve been treating it, has the figure moved since.

Magnesium

Magnesium is the one everybody in this community is already taking, usually for everything, and the evidence is narrower than the reputation.

Let’s cover the measurement problem first, because it shapes the rest of what we will talk about. Most of the body’s magnesium isn’t in the blood at all, so a serum magnesium result is an insensitive way of telling whether somebody is actually short ^[25]. Which is awkward, because it means neither you nor your GP has a straightforward way of answering the one question that would settle whether to even bother. Exactly the sort of gap a confident industry grows in.

On form you’re ahead of the curve though, and the literature agrees with you as far as it goes. Reviews consistently describe magnesium oxide as less well absorbed and more likely to cause diarrhoea, with glycinate, citrate, lactate and malate tending to have better bioavailability, meaning more of what you swallow gets into you, and better gut tolerance, while noting that direct head to head human trials remain limited ^[25]. So, preferring glycinate over oxide is a defensible choice built on absorption and tolerance rather than on a trial that compared the two in people like us. That’s a reasonable basis for picking one off the shelf and a poor basis for telling somebody else theirs is wrong.

On what it actually does, the most reproducible magnesium result is a small reduction in blood pressure, about 2.0 mm Hg systolic and 1.8 mm Hg diastolic pooled across randomised double blind placebo controlled trials ^[26]. The signal turns up at doses of around 300 to 368 mg a day, over longer durations ^[27]. Real, measurable, and a very long way from what gets claimed. For sleep, pooling the work in older adults gave a link towards falling asleep faster, off a small and low quality evidence base though ^[28].

If magnesium settles your gut, your legs or even your sleep, and you can afford it, carry on, it’s cheap and the downside is mostly bowel related (working title of my autobiography). Pick a form you can tolerate, which usually means not oxide, unless constipation is the thing you’re treating. What it won’t do is fix pain, fatigue and joint instability, and the gap between what it has been shown to do and what it gets pushed in marketing as doing, is about as wide as this gets.

Vitamin D

Vitamin D splits cleanly in two, and keeping the halves apart saves a lot of money……. and disappointment.

Correcting a genuine deficiency is absolutely worth doing, and there are good reasons to have it checked, particularly if you aren’t getting out much. What the large trials repeatedly found is that giving vitamin D to adults, who already had enough of it, didn’t deliver broad benefits across fractures, falls, cancer incidence or major cardiovascular events ^[29][30].

So vitamin D is a test first substance rather than a take anyway substance. If your level is low, correct it. If it isn’t, swallowing more of it is unlikely to be the thing that changes much.

Glucosamine, omega-3, ashwagandha and the rest of the drawer

The same logic disposes of most of what’s left, so here’s a quick pass through it.

B12. Correcting a real deficiency matters. Supplementing in people without an overt deficiency is likely to be ineffective for cognition or depressive symptoms ^[31]. An argument for testing rather than an argument against B12. Worth knowing as well that a shortfall here can come out of a medicine cabinet rather than a diet, and proton pump inhibitors, the reflux drugs whose names end in prazole, and metformin are both common in our population.

Multivitamins. Routine supplement use isn’t associated with a mortality benefit in the cohort data, and some supplemental nutrients taken in excess sit somewhere between neutral and harmful ^[32]. A multivitamin is a hedge against a diet you aren’t managing. It isn’t a treatment.

Omega-3. Pooled across the analyses, omega-3 came out with reduced myocardial infarction, which is a heart attack, and some coronary outcomes, with dose dependence in some of them ^[33]. Formulation matters, and the risk of atrial fibrillation, a fast and irregular heart rhythm, rises in some trials and pooled analyses ^[34]. So, not the anti-inflammatory cure all it gets sold as, but not nothing either.

Ashwagandha. Pooled across the randomised trials, ashwagandha came out with reduced anxiety and stress, on an evidence base with high variation between studies and low certainty ^[35]. Promising rather than settled, and the extracts differ from each other enough that a claim made about the whole category is blunter than it looks.

Glucosamine and chondroitin. Straight out of the ingredients list argument we opened with. Cartilage contains them, therefore swallow them, and the swallowing step is the one that was never established. For the sales page to be right, they would have to survive your digestive system, arrive at a joint surface, and be taken up and put to use there. Apply exactly the suspicion you have just learned to apply to collagen.

Protein, and a word on creatine

Here’s the dull answer that outperforms most of the supplement shelf.

Protein supplementation increases strength and fat free mass, during prolonged resistance training, with the benefit plateauing somewhere above total intakes of around 1.6 g per kilogram of body weight a day ^[36]. Two things in that. It works alongside training rather than instead of it, and there’s a ceiling, above which more protein is expensive urine.

Why it matters more here than it does for most people is that rehab is muscle work, and muscle repairs out of amino acids. If you aren’t meeting the baseline requirement in the first place, which happens when appetite is poor, when gut symptoms make protein rich food unappealing, and when cooking on a bad day is beyond anybody, that shortfall is worth correcting before you spend a penny on anything else in this drawer. A shake is a perfectly respectable way of getting there when chewing feels like work.

Creatine is the other one with genuinely strong backing, and it has a post of its own on this site so we won’t re-run it here. In short, creatine improves strength and lean mass during resistance training ^[37]. The body composition effect is small, with fat free mass gains of around 0.8 kg in a large dose response analysis ^[38]. And it works best paired with resistance training rather than taken on its own ^[38]. Which, again, puts the training first and the powder second.

Where we’d put the money

Pulling all of that into something you could act on tonight.

Test before you supplement, and get the figure. Ferritin with transferrin saturation alongside it, vitamin D, B12, and whatever else your history points at. Ask for the actual numbers rather than accepting “normal”, because the reference interval and the threshold at which you’d feel different are two different numbers and only one of them is printed on the letter ^[13][14].

Fix what you find, and change the schedule before you change the product. If iron hasn’t moved in years, alternate day dosing has randomised trials behind it ^[20][21], and a form you can keep down is worth more than a form you abandon ^[22][23].

Meet your protein requirement. Boring, cheap, and it’s the one that works alongside the rehab you’re already doing rather than instead of it ^[36].

Reasonable to try, with modest expectations. Collagen peptides alongside loaded training, which is the setting the tendon results came out of ^[3]. Magnesium in a form you’re more likely to tolerate, for sleep and gut symptoms rather than for pain ^[25][28]. Creatine, taken alongside resistance training ^[37].

Not worth your money. Anything sold to you on the grounds that your tissue is made of it. Anything whose evidence comes off a multi ingredient formula and then gets credited to a single ingredient ^[11]. Anything promising to change the tissue itself, because the instructions decide what gets built and the raw material doesn’t.

The pattern in all of this is that the things with the best evidence are the ones that ask something of you, the training, the protein, the repeat blood test, the awkward conversation about a figure you were told not to worry about, while the things that ask nothing of you except money, are the ones propped up on the ingredients list. That isn’t a coincidence. A product that needs no effort out of you has to be sold on a story instead of a result.

If you’ve spent more on this topic than you’ve ever got back out, that’s an industry that knows precisely who it’s talking to and is very good at it. Knowing how the argument is built is what stops it working on you a second time, and it’s a good deal cheaper than another tub or bottle…………or diarrhoea.

Adam

References

[1] Malfait, F., Castori, M., Francomano, C.A., Giunta, C., Kosho, T. and Byers, P.H. (2020) ‘The Ehlers-Danlos syndromes’, Nature Reviews Disease Primers. https://doi.org/10.1038/s41572-020-0194-9

[2] Pu, S.Y., Huang, Y.L., Pu, C.M., Kang, Y.N., Hoang, K.D., Chen, K.H. et al. (2023) ‘Effects of Oral Collagen for Skin Anti-Aging: A Systematic Review and Meta-Analysis’, Nutrients. https://doi.org/10.3390/nu15092080

[3] Jerger, S., Centner, C., Lauber, B., Seynnes, O., Sohnius, T., Jendricke, P. et al. (2022) ‘Effects of specific collagen peptide supplementation combined with resistance training on Achilles tendon properties’, Scandinavian Journal of Medicine & Science in Sports. https://doi.org/10.1111/sms.14164