A scaphoid nonunion means a fracture in the scaphoid bone has failed to heal, leaving two fragments of bone, sometimes more, that never join back together. Some smaller bones tolerate this well. A fractured toe that never quite unites, for example, rarely causes a lasting problem, because it is not directly loaded when you walk. The scaphoid is different. It sits at the base of the thumb and forms an integral link in the wrist, and every time you put weight through the hand, make a fist, or lift something, an unhealed scaphoid can produce pain.
The reason lies in how the scaphoid works with its neighbours. The forearm has two bones, the radius and the ulna. The wrist itself has eight small carpal bones, three of which form the first row. The scaphoid, together with the lunate and triquetrum, forms the link between the forearm and the rest of the hand. When the scaphoid is fractured and not moving in synchrony with the bones around it, the wrist stops working as a coordinated unit, and that is what turns a bone that has not healed into a problem that causes pain.
When people fall, most instinctively put a hand out flat to break the fall, and the scaphoid is very often the first part of the wrist to make contact with the ground, which is part of why it fractures so frequently. The difficulty comes afterwards, in actually seeing the fracture.
The scaphoid has an unusually curved, undulating shape rather than the simple rectangular outline of many other bones. The name itself comes from the Greek word for boat, describing that curve. For a fracture line to show up clearly on an X-ray, the X-ray beam needs to pass almost exactly parallel to the line of the break. That is straightforward with a bone that has flat, regular edges. It is much harder with a bone that curves the way the scaphoid does. This is why a suspected scaphoid fracture is assessed with dedicated scaphoid views, four X-rays rather than the usual two, each positioning the wrist slightly differently to try to catch the fracture line at the right angle.
Even then, some fractures do not show up on the first set of X-rays. Patients are sometimes asked to return for repeat X-rays a couple of weeks later, because the broken edges of the bone resorb slightly in the short term after a fracture, which can make a previously subtle gap more visible.
Two things need to be in place for any fracture to heal: enough biological capacity to grow new bone, and enough mechanical stability for that new bone to bridge the gap. The scaphoid can be vulnerable on both counts.
Much of the scaphoid's surface is covered in articular cartilage, the smooth lining that allows a joint to move without the underlying bone surfaces wearing against each other. Blood vessels cannot pass through cartilage, so the scaphoid's blood supply enters through a small area at the end of the bone nearer the fingertips, and travels back towards the wrist. This is described as a retrograde blood supply, and it is the single most important fact in understanding this bone. A fracture near the fingertip end, the distal scaphoid, heals extremely well because the blood supply there is excellent. A fracture through the middle, the waist, has a good chance of healing because the blood supply is still reasonable. A fracture towards the wrist end, the proximal pole, effectively has its blood supply cut off, and heals far less reliably without surgery.
Stability matters just as much. The long bones of the forearm, such as the radius and ulna, heal by forming callus: a blood clot converts into soft, immature bone that bridges the fracture as a scaffold, which then gradually hardens into solid bone. The scaphoid, sitting almost entirely within a joint, does not heal this way. Its healing process, intramembranous bone healing, does not tolerate a gap between the fragments, and does not tolerate ongoing movement across the fracture either. If a fracture is not recognised and stabilised promptly, continued movement at the break can be enough on its own to prevent healing.
Two further factors are well established to reduce the chance of healing. Smoking is one: nicotine inhibits bone healing, and in a bone with an already limited blood supply, this matters more than in most other fractures. The other is delay. A fracture treated within around six weeks of injury, even if it was initially missed, still has a good chance of healing with surgery. Left for 18 to 24 months, that chance falls considerably, even with an operation.
Some patients have no symptoms at all. This is why it is common to see patients in their forties presenting with wrist arthritis, only to discover an old scaphoid fracture from an injury in their twenties that was assumed at the time to be a simple sprain.
Where symptoms do occur, pain is typically felt in the anatomical snuffbox, the small hollow on the thumb side of the wrist. The exact location varies with where the fracture sits: a proximal fracture tends to cause pain within the snuffbox and over the back of the wrist, while a distal fracture, nearer the tubercle, tends to cause pain more on the palm side, further towards the fingers.
The pain is usually activity-related rather than constant, particularly once the injury is not recent. Most patients describe the wrist as comfortable at rest, with a sharp burst of pain triggered by gripping, pushing through the hand, lifting, or pushing up from a chair, which then settles once the hand is rested. Movement is often lost at the extremes of wrist motion, and particularly with ulnar and radial deviation, moving the wrist from side to side. Grip strength often falls, generally because pain discourages normal use of the hand rather than because the muscles themselves are affected.
It is worth noting that a fracture does not only heal completely or not at all. Sometimes the fragments join with fibrous or scar tissue rather than solid bone, known as a fibrous union, which can allow the wrist to function reasonably well even though the two ends are not fully and normally healed. In other cases, the fragments never develop even this loose connection and continue to move independently.
On examination, findings that point towards a nonunion rather than a simple sprain include tenderness in the anatomical snuffbox, tenderness over the scaphoid tubercle on the palm side, pain when the scaphoid is axially loaded (pushing the thumb in towards the wrist), and pain on wrist movement, sometimes alongside subtle swelling, though visible bruising is often absent.
X-ray is the first-line investigation: it is quick, widely available, and identifies most fractures. If the X-rays are normal but suspicion remains high, MRI is generally the next step, rather than CT. NICE guideline NG38 advises that MRI should be considered as first-line imaging in people with a suspected scaphoid fracture following a thorough clinical examination, reflecting how often the initial X-ray misses these injuries. This is a deliberate choice. MRI takes multiple slices from different angles to build a three-dimensional picture, and its particular strength is detecting oedema, fluid within the bone that behaves like a bruise and can reveal a subtle fracture invisible on X-ray. It carries no radiation, which matters because CT involves a meaningfully higher radiation dose than a standard X-ray and would not be justified purely to make an initial diagnosis. MRI can also identify a different injury, a scapholunate ligament tear, which causes very similar pain but would not show up on a CT scan.
CT comes into its own once a nonunion is established and an operation is being planned, rather than at the diagnostic stage. It shows the number of fragments, their alignment, and the presence of any gap in detail that guides the surgical approach directly: whether fixation should come from the palm side or the back of the hand, whether the approach can be minimally invasive or needs to be open, and whether a screw or a plate is the right choice.
The outcome depends heavily on where the fracture sits. A distal tubercle fracture that fails to unite may never cause any problem and can be found incidentally on an X-ray taken for something else entirely, years later. A proximal pole fracture is a different matter: with its blood supply effectively cut off, the fragment can undergo avascular necrosis, where the bone loses its blood supply and gradually dies. This may cause no symptoms for some time, but as the fragment loses substance, it can begin to cause pain and premature wear against the neighbouring bone.
A waist fracture that fails to unite tends to develop a characteristic pattern: the distal part of the scaphoid flexes one way while the proximal part extends the other way, producing what is known as a humpback deformity. Instead of its normal shape, the bone develops a collapsed, angulated profile, and because the two fragments are now moving incoherently rather than as one, this accelerates wear elsewhere in the wrist.
Over years, an unhealed scaphoid can lead to a specific and fairly predictable pattern of wrist arthritis, beginning at the radial styloid at the base of the thumb, then progressing through the joint between the scaphoid and capitate, and in advanced cases eventually reaching the joint between the lunate and radius. This is known as SNAC wrist, scaphoid nonunion advanced collapse, and it is a large enough subject to deserve its own detailed article, which we cover separately. Having arthritis visible on an X-ray does not automatically mean it will cause pain, and some fracture patterns, such as a tubercle fracture, are unlikely to ever progress this way. But this natural history is precisely why an old, dismissed wrist injury is worth having properly assessed rather than assumed to be a simple sprain that happened to leave some stiffness behind.
The decision depends on three things above all: where the fracture is, how long it has been present, and whether the patient has already had treatment.
A proximal pole fracture that has not healed after around three months is unlikely to unite on its own; the risk of this particular pattern failing to heal without surgery is often considered to be as high as 90% in clinical practice, though published series vary on the exact figure. Given that a proximal pole fracture would also need an extended period in a cast, often eight to ten weeks rather than the more typical six, with no guarantee of success at the end of it, there is often a low threshold for recommending surgery in this situation from the outset. A minimally invasive fixation reduces the two fragments together, removing any gap, while disturbing the surrounding blood supply as little as possible.
A waist fracture that is well aligned, has no gap, and is caught early behaves very differently, and the majority of these heal well with cast treatment alone. There is a long-running debate in hand surgery about whether the cast should include the thumb. The evidence suggests the healing outcome is much the same either way; leaving the thumb free allows better use of the hand during recovery, while including it can offer some patients extra reassurance, particularly if there is concern about them protecting the wrist adequately otherwise.
A distal tubercle fracture, where the blood supply is excellent and a humpback deformity is unlikely to develop, generally does very well with a shorter period of immobilisation, often around four weeks rather than six to eight.
Occupation and lifestyle also factor into the decision even where a fracture would likely heal in a cast. A professional athlete or someone needing a fast return to activity may still opt for a small operation, a screw placed through a very small opening in the palm, purely because it removes the need for cast immobilisation entirely and allows a quicker return to driving and sport, even though it does not necessarily change the long-term outcome compared with successful cast treatment.
Smoking and delay beyond twelve months both reduce the likelihood of a good outcome, whichever treatment path is chosen. Once a patient has developed established arthritis, fixing the fracture no longer addresses the actual source of pain, and the conversation shifts towards the salvage options used for arthritis rather than reconstruction of the original fracture.
Where there is any uncertainty about whether arthritis has already developed, particularly in a longstanding nonunion where the X-rays are not conclusive, wrist arthroscopy, examining the joint directly with a small camera, is often used before committing to fixation. It is considerably more definitive than MRI for confirming whether the joint surfaces remain intact.
Arthroscopy is increasingly used as a treatment tool as well as a diagnostic one. Where no arthritis is found, bone graft can sometimes be placed and the fracture fixed under arthroscopic guidance, typically supplemented with wires rather than a plate. The particular advantage of this approach is that it avoids disturbing the blood vessels around the scaphoid on the way to the fracture, which matters in a bone that is already vulnerable on that front.
The CT scan drives the surgical plan, and several specific questions are worked through in sequence.
The first is which fragment is smaller. The convention is to fix the smaller fragment to the larger one, and the surgical approach, from the palm side or the back of the hand, follows from wherever the smaller fragment happens to sit.
The second is how many fragments there are. Two clean fragments can usually be compressed together with a single headless screw embedded within the bone, which draws the fragments together and reduces movement across the fracture. Where the bone is broken into several pieces, compressing them together can actually push fragments apart rather than uniting them, so a plate with several smaller screws, acting as a scaffold, is generally more appropriate.
The third is whether a humpback deformity is present. Correcting this means separating the collapsed fragments and lifting the bone back into its proper shape, which leaves a gap that needs to be filled with a structural wedge of bone, usually taken from the back of the wrist. This graft is not simply filling space: it also brings biological healing factors with it.
The fourth is whether the proximal pole still appears to have a viable blood supply. If it does, a standard bone graft aimed at boosting the local biology is generally sufficient. If the proximal pole appears to have lost its blood supply (avascular necrosis), a vascularised bone graft is used instead, transferring a piece of bone, commonly from the back of the distal radius, together with its blood vessel still attached, so that the graft brings a genuine new blood supply into the area rather than biology alone.
The CT scan is also used to check for arthritis. If arthritis is already present, fixation no longer serves its purpose, since the aim of fixing a nonunion is to prevent arthritis developing in the first place; if it has already developed, the patient is likely to have ongoing pain regardless of whether the fracture itself goes on to heal. In that situation, treatment shifts to the salvage options aimed at the arthritis. Previous unsuccessful surgery also changes the plan, generally pushing towards a more robust combination of plate fixation and vascularised graft, since each prior operation reduces the bone stock and further disturbs the local blood supply.
Bone graft can come from two sources. Autograft is bone taken from the patient's own body, and has the advantage of being living tissue that carries its own biological healing factors, including bone morphogenetic proteins that actively encourage new bone formation. Allograft, taken from a donor or from processed bovine bone, provides a structural scaffold without that biological contribution, and avoids the need for a second surgical site, though it is generally considered less effective than the patient's own bone where biology is the main concern.
Which is used, and where from, depends on what the graft needs to achieve. If the fracture simply needs a biological boost, cancellous bone (the softer inner part of bone) taken from the wrist itself is usually sufficient, and avoids opening a second wound elsewhere in the body. If structural support is also needed, for a humpback deformity for example, a wedge of bone including the harder outer cortex is required. This can often still be taken from the back of the wrist, though if a larger graft is needed, the iliac crest at the hip provides a bigger and richer source at the cost of typically being more uncomfortable afterwards than the wrist donor site. For that reason, most hand surgeons prefer to use the wrist where the graft required is small enough to allow it.
Where the proximal pole has already lost its blood supply, structure and biology from a graft are not enough on their own, and a vascularised graft, bone transferred together with its own blood vessel, is used to restore an actual blood supply to the area.
As with any operation, there are risks that should be discussed individually with your surgeon, though most are uncommon:
Surgery is performed as a day-case procedure: patients come in, have the operation, and go home the same day. The length of the operation depends on complexity, from around half an hour for a percutaneous screw fixation to up to two hours for an open reconstruction with bone grafting. Most operations are performed under a regional block, numbing the arm, with the option of light sedation for patients who would rather not be aware of the procedure; general anaesthetic is available where needed.
Time in plaster varies considerably, from none at all for some minimally invasive fixations, up to around six weeks in a cast followed by a further period in a splint for more complex reconstructions. Driving is not possible while in a cast, and once out of a cast, is generally reasonable from two to four weeks, provided the wrist can control the vehicle safely. Return to work depends on the nature of the job: desk-based work is often possible within a couple of weeks even in a cast, while heavy manual work, such as building work, can require up to three months away.
Healing is monitored with X-rays at intervals such as six weeks, three months, and six months. A CT scan is not usually required in the first six months, but may be used after that point to confirm healing definitively, particularly where metalwork makes the X-ray appearance harder to interpret. It can take up to twelve months to be fully confident that the fracture has healed, and while many patients begin to feel that the wrist has returned to normal from around six months, that timeline varies from person to person.
"It is too late to do anything." Even a scaphoid fracture that has gone untreated for months can usually still be assessed for treatment. Location, the presence of a gap, the number of fragments, and factors such as smoking or diabetes all affect the likelihood of success, but the presence of an old, unhealed fracture is rarely a reason on its own to rule out treatment.
"If there is arthritis, nothing can be done." Once arthritis has developed, fixing the original fracture becomes less relevant, because the ongoing pain is now coming from the arthritic joint rather than the unhealed bone itself. The focus shifts from mending the fracture to managing the arthritis directly, and there are several established ways of doing that, covered in more depth in our separate article on SNAC wrist.
Confusion about bone graft. Patients are often unclear on why a graft is needed and where it comes from. It comes down to two simple questions: is the graft needed to boost biology, to provide structure, or both, and can it be taken from the patient's own wrist without needing a larger donor site elsewhere.
Can a scaphoid nonunion heal without surgery?
It depends heavily on where the fracture sits. A well-aligned waist fracture caught early often heals well in a cast alone. A proximal pole fracture is far less likely to heal without surgery.
Is it really ever "too late" to treat a scaphoid nonunion?
Rarely. Even fractures that have gone unrecognised for months can usually still be assessed and treated, though the chances of success reduce the longer treatment is delayed, particularly beyond a year, and with smoking or previous failed surgery.
Does smoking really make that much difference?
Yes. A 2021 systematic review and meta-analysis of over 400,000 patients, published in eClinicalMedicine, found that smokers had more than double the risk of nonunion compared with non-smokers after fracture treatment. Nicotine inhibits bone healing, and in a bone with an already limited blood supply, this can be the difference between a fracture uniting and progressing to nonunion.
How long does recovery take?
This varies with the complexity of the surgery. Many patients return to desk-based work within a couple of weeks, while full confidence that the bone has healed can take up to twelve months.
What is a humpback deformity?
It is the collapsed, angulated shape a scaphoid can develop when a waist fracture fails to unite, with one part of the bone flexing and the other extending. Correcting it surgically usually requires a structural bone graft to restore the bone's original length and shape.
A scaphoid nonunion found soon after an injury, or discovered incidentally many years later, both warrant the same first step: proper imaging and assessment by a specialist hand and wrist surgeon before any decision is made about treatment. iiS Health, The Cheshire Hand Clinic, provides this assessment, including on-site X-ray, with CT or MRI arranged where required.
.png)