What's Actually Going On
The radius carries roughly 80% of the load across the wrist, and its lower end is metaphyseal bone: a thin cortical shell wrapped around a lattice of trabecular bone. That construction is why it is the first thing to fail in a fall onto an outstretched hand, and why it fails at low energy once bone density drops.
The classic Colles pattern is a break that does not enter the joint, with the fragment tipped backwards. The division that actually matters clinically is not the eponym but whether the fracture crosses the joint surface. Complete articular fractures (AO-OTA type C) carry a significantly higher rate of acute carpal tunnel syndrome and a significantly higher complication rate after plating, and joint-surface incongruence is the predictor of later arthritis.
Why the median nerve is the structure that matters here. The carpal tunnel sits immediately beyond the fracture, roofed by a ligament that does not stretch. Bleeding, swelling and displaced fragments raise the pressure inside a compartment that cannot expand. That is acute carpal tunnel syndrome, and it is not a same-day event. The median onset is one week, which is after the fracture clinic has finished and around the time a therapist first sees the patient.
What rehabilitation is actually treating. Not the bone. The bone heals on its own schedule. Rehabilitation treats the secondary cost of protecting it: capsular and muscle-tendon stiffness from a wrist held still, plus the loss of grip that comes from not using the hand. That distinction runs through the whole evidence base, and it explains why every intervention with a reliable effect works by shortening or softening the immobilisation rather than by adding exercise on top of it.
How to Identify It
This diagnosis is radiographic. It is not made or excluded by a clinical special test, and this card does not pretend otherwise.
| Test | What it tells you | Accuracy |
|---|
| Radiography (PA, lateral, oblique) | The diagnosis itself | Human readers detect distal radius fractures at Sn 0.95 (95% CI 0.91–0.97) against 0.71 for the scaphoid in the same review against the same reference standard (Suen 2024, PMID 38981178). Carried across from the scaphoid card's evidence base, not this card's sweep |
| Median nerve sensory screen | Acute carpal tunnel syndrome | Sn/Sp: data unavailable Justified by incidence (4.3%) and timing (median 1 week), not by a published likelihood ratio |
| Active thumb IP extension | Extensor pollicis longus integrity | Sn/Sp: data unavailable Late screen; tendon complications 4.7% after volar plating |
Honest gap, stated rather than filled. The 80-paper evidence sweep behind this card carries no sensitivity or specificity data for any clinical decision rule in wrist fracture. Those cells say so instead of being filled in from general clinical knowledge.
What to actually examine: the distal radius and the distal radioulnar joint, but also the anatomical snuffbox and scaphoid tubercle, because the scaphoid is the neighbouring bone that gets missed. Check shoulder, elbow and finger range every time, because that is the range most often lost while the wrist is protected.
The Debate
Supervised therapy versus a home programme: the field reversed in 2026
Then: three syntheses found no difference. Five of seven trials showed nothing (Valdes 2014), 13 trials were judged insufficient (Bruder 2017), and 15 studies found no difference on any outcome with GRADE rated very low on every single analysis (Soares 2023), whose authors said outright that the evidence was insufficient to support even their own null result.
Now: 13 RCTs searched to April 2025, using formal risk-of-bias and certainty assessment, found supervised physical therapy superior at 6 weeks with moderate certainty, plus a significant effect in the over-65s and a dose-response (Gutiérrez-Espinoza 2026).
Which to follow: the 2026 analysis. The older nulls were underpowered rather than negative, and they said so themselves. Update the conclusion rather than averaging it.
Does the early-mobilisation benefit last a year?
Pooled across 5 RCTs: at one year, not one variable differed between early and delayed motion (Gutiérrez-Espinoza 2021).
One RCT disagrees: 10.2° of extension/flexion and 7.9 Mayo Wrist points still present at one year (Quadlbauer 2022, N=116).
Why both can be right: they are not testing the same thing. The pooled trials define "early" as within 2 weeks against 5 to 6 weeks. Quadlbauer randomised day-1 mobilisation in a removable splint against a non-removable five-week cast, a far wider contrast. A small early-versus-slightly-later difference washing out is compatible with a large immediate-versus-immobilised difference persisting.
The same research team, the opposite headline, six years apart
Bruder 2011 concluded that exercise reduces impairment and improves activity after upper limb fracture, across 13 trials. Bruder 2017, same first author, same team, same journal, concluded that prescribed exercise programmes may not be effective, across 22 trials.
What changed: nine more trials and a reporting-quality assessment which found fewer than 40% of them described well enough to replicate. What did not change: both reviews agree that starting exercise early with reduced immobilisation improves activity. The reversal is about generic exercise volume, not about the timing decision.
Honest Limitations
The trials studied the uncomplicated patient; the clinic sees the complicated one
The Cochrane review states that with few exceptions its 26 trials excluded people with serious fracture or treatment-related complications, and older people with comorbidities and poor overall function, precisely because those people would have needed more intensive treatment. So the reassuring finding that home exercise equals supervised therapy was established in exactly the population least likely to be referred to a therapist in the first place.
The exercise prescriptions are not reproducible
Fewer than 40% of trials described their programme sufficiently to replicate, and across all 80 papers reviewed here exactly one specifies a quantified loading parameter. A clinician asking this literature "how many sets, at what load, progressed how" gets no answer. Saying so is more useful than importing a number from elsewhere and calling it evidence.
The clinical-importance test is available and is applied unevenly
Across 80 papers there are two mentions of a minimal clinically important difference. The one research group that applied the test applied it to the surgical comparison and concluded those statistically significant differences were not clinically important. The same group's rehabilitation findings of comparable size are reported as statistically significant without that test being run. This is an observation about a yardstick being used asymmetrically, not an accusation of bias, and it is why no threshold value is quoted anywhere on this page.
The evidence base does not describe the population it studied
A review of 77 rehabilitation trials published between 1987 and 2021 found sex and gender adequately considered in six of them, while the cohorts here run 73% to 96% female. Nothing in the prescription changes, but confidence in any subgroup reasoning should be low.
The Nuance
Surgery versus a cast, and the honest version of it. In adults over 60, a volar plate was statistically superior to a cast at one year on wrist function, disability, grip and quality of life. The authors of that meta-analysis then ran the clinical-importance test on their own result and concluded the differences were not minimally clinically important, and that both managements are equally effective (Gutiérrez-Espinoza 2022). The VOLCON randomised trial in patients over 65 found complications of 20.9% after surgery against 16.6% without, p=0.78 (Thorninger 2022).
In adults aged 18 to 64 the trade is clearer, and it is a trade rather than a win. Across 34,184 fractures, surgery produced more stiffness at one year (202.8 versus 123.4 per 1,000) and far fewer secondary procedures (8.7% versus 43%), with carpal tunnel release the most common secondary procedure (DeGeorge 2021). Stiffness, not failure of the bone to heal, is the one-year problem, and stiffness is the thing rehabilitation exists to prevent.
The sentinel-fracture question, downgraded on the evidence. A distal radius fracture in an older adult is an established marker of future functional decline and hip fracture, and the combined hip-and-wrist fracture population is overwhelmingly female with longer hospital stays. But the specific claim that these patients have measurably weaker legs does not hold up yet: a review of 19 studies and 1,835 participants could not pool its results numerically, because 12 studies measured strength in 10 different ways and 18 measured balance in 14 different ways, leaving the findings conflicting and inconsistent (Forde 2023). Screening an older adult for falls risk after this fracture is justified by the epidemiology. Asserting a measured lower-limb deficit is not.
