What Works
The strongest recommendations on this page are things not to do. That is genuinely what the evidence says, and it is worth more than a protocol invented to fill the gap.
Do nothing, while the hand still works Strong
No treatment is indicated for a lump without a bend, or for a bend that is not limiting you. In a population followed for seven years, 21.5% got worse and 6.5% actually improved. Over five years, 11 to 16% of hands were stable or improved. Doing something to a hand that still works can turn a stable finding into a complication.
After surgery, do not splint routinely Strong
A trial across five hospitals randomised 154 patients after surgery and followed 96% of them for a full year. Night splinting made no difference to hand function, to how straight the finger was, or to how satisfied patients were. The trial authors recommend against giving it to everyone.
A procedure to release the cord, once the bend genuinely limits you Strong
Surgery, a needle release, or a collagenase injection. All three straighten the finger about as well as each other. What differs is the trade: surgery lasts longest before it comes back and carries the highest risk of nerve or vessel injury; the injection and needle release are gentler and quicker to recover from, and it returns sooner. Serious complications did not differ between them.
Exercise Prescription
Night splint after a collagenase injection, middle joint only Moderate
A 2025 trial of 69 patients found a night splint left roughly 15 to 17 degrees less bend at four months, but only at the middle finger joint. At the knuckle joint it made no difference at any point. One trial, so treat it as promising rather than settled.
Movement and exercise after a procedure Moderate
Guideline-backed. Dose unproven. The one thing worth protecting is your bend: it is common to chase straightening so hard that full fist gets lost, and you need both.
Tier 3 — Emerging, and not routine care
An injected drug for early disease. Emerging
A trial programme is testing injecting an anti-inflammatory biologic drug directly into early lumps. It does make lumps smaller and softer on a scan at twelve and eighteen months. Here is the catch, and it matters: the first trial missed its own main target in 28 people, and nobody has yet published that any of this keeps a finger straight. Softer lumps on a scan is a reasonable step toward preventing a bend. It is not the same thing as preventing one.
Steroid injection into a painful lump. Emerging
Appears to soften lumps and ease pain, on poorly designed studies. Reasonable if a lump genuinely hurts. It is pain relief, not disease control.
What Doesn't Work
- Stretching, massage, or trying to break the cord down. No study we found even tests this, for benefit or for harm. It persists because it feels like something ought to loosen it, and because patients ask for it.
- Routine night splinting after surgery. Tested properly in 154 patients and found to make no difference at one year. It persists because it is intuitive and short-term gains look convincing in clinic before they wash out.
- Splinting or therapy to stop a lump becoming a bend. Nothing supports it, and some of those "successes" were never going to progress anyway.
- Radiation treatment for early disease, as established care. Across 770 treated hands, reported improvement ranged from 0% to 56% and worsening from 2% to 86%. Numbers that wide are not a measurement.
- Scans to predict whether it will progress. Directly contested: one study found the scan appearance did not predict lump growth at all.
Red Flags
Most Dupuytren's is slow, painless and safe to leave alone. These are the exceptions.
- A lump that is painful, growing quickly, soft, movable, or sitting on the back of the hand. Dupuytren's lumps are firm, fixed in place, and sit in line with a finger. An unusual lump needs looking at, not watching.
- You can no longer fully bend the finger. This condition takes away straightening and never takes away bending. Losing bend means something else is going on.
- After any procedure: numbness, pins and needles, or a finger that looks pale or dusky or feels cold. Contact your surgical team promptly.
- A hand that is hot, red, swollen and severely painful. That is not this condition. It needs assessing for infection or an inflammatory cause.
- A finger bending fast, in a younger person, with lumps on the knuckles or soles of the feet and a strong family history. Worth an earlier hand surgery opinion.
Return to Training
Before any procedure there is no restriction to apply and none is evidence-based. Gripping and lifting do not accelerate this disease. If a lump is sore under a bar, that is a comfort problem with a comfort fix: padded grips, straps, handles or a trap bar. Tell people this explicitly, because many quietly stop training out of fear.
After a procedure, the wound is the constraint. No loaded gripping for the first two weeks, progressive return over six.
- Wound fully healed, no drainage, no signs of infection
- Full active bend restored — a complete fist. This is the half most people neglect
- Straightening holding steady, with no progressive loss over two consecutive weeks
- Grip strength around 80% of the other hand before loaded gripping
- Three sets of ten at a previous working weight on a grip-heavy lift, with no sharp scar pain
- No swelling or pain lasting beyond 24 hours after a session
- Cleared by the surgical team where a skin graft or a complication was involved
Put your palm flat on a table and press down. If it will not lie flat, that is the tabletop test, and it is the same one a hand surgeon uses. Now check the other half: make a full fist. With Dupuytren's you still can. If you cannot bend the finger, this is not what you have.
Under the skin of your palm there is a flat sheet of tough tissue that anchors the skin so your grip does not slide. In Dupuytren's, cells inside that sheet start laying down new fibres and pulling them tight, like a rope being rewoven a little shorter every month.
That is why stretching does nothing. You are not fighting a tight muscle that will eventually let go. You are fighting a rope that is actively shortening itself. It is also why an operation works: someone has to cut the rope.
- What this actually is: it is not arthritis and it is not your tendon, which is exactly why the finger still bends perfectly well and just will not straighten.
- The myth that won't die: that using your hands caused it, or makes it worse. It is mostly in your genes, and gripping, lifting and working do not speed it up.
- Start here: stop treating it as an emergency. Over seven years about one in five people got worse, and a small number actually got better on their own.
Conviction Moderate
The strongest claims on this page are negative ones, and that is what makes them useful. Negative findings from properly sized trials are the most reliable material this topic offers, and they are the ones you can act on today.
Per claim: routine splinting after surgery does not help — moderate to high. No disease-changing role for hand therapy — moderate to high. Night splint after collagenase for the middle joint — low to moderate. Progression is slow and sometimes reverses — moderate. Radiation for early disease — low. The injected drug changing the disease — low. Accuracy of the hand tests — no evidence exists.
What would change my mind about the splint after collagenase
A trial of at least 200 people, middle-joint bends only, with wear time measured by a sensor rather than self-reported, followed to twelve months with a pre-agreed definition of recurrence. If it reproduced the same gap at a year with verified wear, this moves from a suggestion to a genuine protocol. If the gap closed by twelve months, as happened in the surgical trial, the 2025 result gets reclassified as a short-term effect that washes out.
What would change my mind about the injected drug
A large trial of at least 400 people with early disease, followed at least three years, whose main measure is whether a finger actually develops a bend rather than how hard or large the lump looks on a scan. That single design change would settle this in either direction. Until it reports, "there is now a drug for early Dupuytren's" is not a supportable statement.
Most hand conditions get treated with confident advice that nobody ever tested. We check first.
Join The Verdict — freeThe Full Picture — Anatomy, Diagnosis & Evidence
What's Actually Going On
The palmar aponeurosis is a flat fibrous sheet lying between the skin of your palm and the flexor tendons underneath. Its job is to anchor skin to the deeper hand so your grip has purchase.
In Dupuytren's, cells within that sheet called myofibroblasts multiply and lay down collagen. The important detail is that these cells are contractile. The tissue does not merely thicken, it actively shortens. First a firm nodule appears, usually in line with the ring or small finger. Later a rope-like cord matures, and that is what produces the fixed bend.
The clinical signature falls straight out of the anatomy. Because the flexor tendons themselves are untouched, the finger still bends normally under its own power. What gets lost is the ability to straighten it. That single pairing is the diagnosis.
The genetics are substantial. A study pooling 11,320 people with the condition against 47,023 without it found 85 significant genetic markers across 56 locations, explaining somewhere between 13% and 38% of why some people get it. Family history is one of only two things that independently predicted getting worse in a seven-year study.
How to Identify It
- A firm lump or rope-like cord in the palm, in line with a finger, fixed to the tissue underneath rather than sliding freely
- Puckering or dimpling of the skin directly over it
- Loss of straightening, with bending fully preserved the key pairing
- Palm will not lie flat on a table (tabletop test) Sn: DATA UNAVAILABLE Sp: DATA UNAVAILABLE
- Cord palpation along the finger line Sn: DATA UNAVAILABLE Sp: DATA UNAVAILABLE
- Often both hands, often a family history, more common with diabetes
Here is why that turns out not to matter much. In sixteen years of general practice records, the diagnosis was correct at the very first appointment 93% of the time, and 73% of patients needed only one visit. The tests have no published accuracy because nobody has needed to study them, not because the diagnosis is unreliable. This is a condition that non-specialists recognise on sight.
What it is not. Trigger finger catches and locks as you bend, and its lump moves with the tendon; the Dupuytren lump stays put. A flexor tendon injury takes away bending, which this never does. A ganglion is soft, squashy and mobile. An ulnar nerve problem claws the same two fingers but comes with numbness and wasting, and the fingers can still be straightened passively. Diabetic stiff hand gives waxy skin and generalised stiffness everywhere, with no discrete cord. Worth knowing: Dupuytren's, trigger finger and carpal tunnel cluster together in diabetes, so finding one is a reason to check for the others.
The Debate
No physical therapy guideline exists for this condition. The nearest authority is a European multidisciplinary guideline that is now thirteen years old, and, notably, contains no non-surgical treatments at all.
Honest Limitations
Nobody agrees what "it came back" means
Recurrence after the injection is quoted at 25.8% against 9.3% for surgery. But a review of 79 studies found a wide range of definitions of recurrence and rated every single study at high risk of bias, and twenty-four hand surgeons from seventeen countries once held an international conference for the sole purpose of agreeing what the word should mean. Part of that gap is real biology and part is that the two sets of studies simply counted differently. Trust the direction, not the size.
Trial compliance is not real-life compliance
In the splint trial, inside a monitored study, the number of people wearing it regularly fell from about 4 in 5 at one month to about half by four months. A four-month nightly splint prescription is realistically a two-month splint, and that is the generous reading, because people were reporting on themselves.
The most persuasive protocol in this literature has no control group
One study reported just 7% recurrence at nearly five years using surgery plus a 24-week night splint plus home exercise, with both recurrences in the patients who stopped wearing the splint early. That is exactly the shape of result that gets a protocol adopted worldwide. It is 30 people, with nothing to compare against, and abandoning an uncomfortable splint is at least as likely to be a sign of worse disease as a cause of it. The study that did have a comparison group found nothing.
The Nuance
This is one of the few conditions where "conservative management" does not mean conservative treatment. Once a finger is genuinely bent, nothing non-surgical straightens it. The real choice is whether to intervene at all, and then which procedure.
| Surgery | Needle release | Collagenase injection | |
|---|---|---|---|
| Straightening | Best on paper, but the difference is not enough to matter | Equal to the injection | Leaves more residual bend than surgery |
| Comes back | Latest — 9.3% | Sooner than surgery | Soonest — 25.8% |
| Serious complications | Highest — 7.3%, most nerve and vessel injuries | Skin tears | Lowest serious, most minor. No difference in major complications |
| Burden | Operation, longest recovery | Office-based, fastest | Injection, repeatable without stacking risk |
All three straighten the finger about as well as each other, and the real decision is what you are willing to trade. Surgery buys the longest time before it returns and charges a genuinely higher risk of nerve or vessel injury for it. The injection and the needle release are gentler and quicker, and it comes back sooner. The injection can be repeated on a recurrence without stacking up risk, which matters more than it sounds for a condition that recurs.
None of that is a reason to intervene on a hand that still works. In general practice, roughly a third of patients are simply watched, and fewer than half are ever referred onward. That is not neglect. Given that some of this disease is stable and a little of it reverses, it is good medicine.
Sources
- Jerosch-Herold C, et al. 2011, BMC Musculoskeletal Disorders (PMID 21693044) — Multi-centre randomised trial, 154 patients, 96% followed to 12 months. Night splinting after surgery null on function, straightness and satisfaction. Strong
- HANDGUIDE study group, 2013, Plastic and Reconstructive Surgery (PMID 24281643) — European consensus, four rounds, 39 experts including hand therapists. No non-surgical treatments included in the guideline. Moderate
- 2025, Journal of Hand Surgery (American) (PMID 40047775) — Randomised trial, 69 completed, 119 joints. Night splint after collagenase helps the middle joint only; no knuckle-joint effect at any point. Moderate
- Broekstra DC, et al. 2022, Plastic and Reconstructive Surgery (PMID 35404338) — Prospective study, 258 people, 17,645 measurements over 5 years. No traditional risk factor predicted progression. Moderate
- Lanting R, et al. 2021, Bone & Joint Journal (PMID 33789472) — Population study, 398 re-examined at 7 years. 21.5% progressed, 6.5% regressed. Moderate
- 2025, Journal of Hand Surgery (European) (PMID 40391547) — Individual patient data meta-analysis, 15 studies. Correction comparable across all three procedures; major complications did not differ. Strong
- 2025, Family Practice (PMID 39566072) — 16-year general practice records. Diagnosis correct at first contact in 93%; a third managed by watchful waiting. Moderate
- Rodrigues JN, et al. 2016, BMC Musculoskeletal Disorders (PMID 27526686) — Review of non-surgical treatment for early disease, 26 studies, all at the lowest two evidence grades. Moderate
- Krefter C, et al. 2017, Journal of Hand Surgery (European) (PMID 28490266) — Review of radiation therapy, 770 hands. Improvement 0–56%, worsening 2–86%. "An unproven treatment." Moderate
- Nanchahal J, et al. 2018, EBioMedicine (PMID 29983350) — Early-phase drug trial, 28 people. Missed its own main target; industry co-funded. Low
- Riesmeijer SA, et al. 2024, Nature Communications (PMID 38172110) — Genetic study, 11,320 cases against 47,023 controls. 85 markers, 56 locations. Strong
- 2024, Occupational and Environmental Medicine (PMID 39317441) — Review of 15 studies, formally graded. Hand-transmitted vibration doubles the odds of developing it. Moderate
Full evidence synthesis: 42 sources, 41 anchored to indexed identifiers. This page is educational self-management guidance, not personalised medical treatment.