Scaphoid fracture

Key points

This page covers breaks of the scaphoid, a carpal bone (one of the small bones of the wrist) on the thumb side¹.
A 2020 review states that X-rays detect no more than 70% of scaphoid fractures².
One randomised trial (a study that assigns people to groups by chance) compared early surgery with a cast in 439 adults³. Their waist fractures had shifted 2 mm or less, and a patient-rated wrist score at one year did not differ significantly³.
In 102 symptomatic patients with a nonunion (a broken bone that fails to join and heal), osteoarthritis (wear of joint cartilage) was present in every nonunion ten or more years old⁴. A 1993 appraisal of the natural history studies published by then concluded that their bias ran toward a worse outcome⁵.
A 2019 systematic review (a study that searches for and sums up the research on a question) pooled eleven studies of athletes⁶. In it, 90% of 77 managed without surgery and 98% of 83 fixed surgically returned to sport⁶. Its authors rated the studies’ quality as lower than in similar reviews of other fractures⁶.

What it is

An open anatomy textbook defines a fracture simply as a broken bone¹. The same textbook lists the scaphoid as the first carpal bone on the thumb side of the row next to the forearm¹. The scaphoid page covers the bone itself, including its blood supply.

Two labelled line drawings of the bones of a hand and wrist, one seen from the front and one from the back. At the base of each hand, above the ends of the radius and ulna, sit the eight carpal bones in two rows. The row next to the forearm is labelled scaphoid, lunate, triquetrum and pisiform; the row next to the palm is labelled trapezium, trapezoid, capitate and hamate. Beyond them are the five numbered metacarpals of the palm and the phalanges of the thumb and fingers. A small inset marks the carpals, metacarpals and phalanges on a whole hand. — Bones of the wrist and hand, front and backBetts JG, Young KA, Wise JA, et al. Anatomy and Physiology, Figure 8.7. OpenStax, 2013. Access for free at openstax.org. (source, CC BY 4.0). Resized from 1954 to 1400 pixels wide; nothing else changed.

How studies define the break

The protocol of SWIFFT, a trial described below, required a break that plain X-rays showed clearly⁷. The break also had to cross both outer walls of the bone’s waist and be shifted 2 mm or less⁷. A 2026 systematic review looked at 58 studies of suspected scaphoid fractures⁸. Nearly half gave no definition of a fracture on MRI, a magnetic scan, and its authors called for an agreed one⁸.

Where the bone breaks

Proximal means nearer the centre of the body, and distal means farther from it. In scans of 43 hand specimens, the centre of the waist lay in the mid-portion of the bone, and the authors called the waist narrow⁹. A 2020 review states that fractures of the proximal pole, the end nearer the body, lack a common definition². Since breaks nearer that end are less likely to join, it notes a proposal to use the bone’s proximal fifth². The SWIFFT protocol describes the proximal pole the same way⁷.

The 2020 review states that fractures of the waist make up two-thirds of scaphoid fractures². Most of those, 60 to 85%, are not displaced². In a displaced fracture, the bone ends have moved out of line. The review puts 25% in the distal third and 5 to 10% in the proximal third².

Fractures that first X-rays do not show

The 2020 review states that X-rays detect no more than 70% of scaphoid fractures². Studies have counted fractures in wrists whose first X-rays looked normal^10,11.

Group studied	Share with a scaphoid fracture
1,685 wrist injuries with normal first X-rays, in a systematic review of eight studies	9.0%¹⁰
1,989 patients at eight UK centres who had an early MRI scan for a suspected fracture	12.9% (256 patients)¹¹

The 2020 review puts the share at 4 to 20%².

Who it affects

A systematic review of 42 epidemiology studies found that adult scaphoid fractures occur mostly in males, with a peak at ages 20 to 29¹². Females peaked earlier, at ages 10 to 19¹². Its authors rated the evidence on who breaks the bone as low¹².

The 2020 review describes the typical injury as a fall in sport onto an outstretched hand that bent the wrist back². An anatomy textbook says the force of such a fall travels up the limb¹. It may break the humerus (upper arm bone), the radius (the forearm bone on the thumb side) or the scaphoid¹. The distal radius fracture page covers the break at the wrist end of the radius.

A stress fracture is a small crack in a bone from repeated loading. A case report is a report on a single patient. A 2016 case report, which also reviewed the literature, states that scaphoid stress fractures are very rare¹³. It adds that the peer-reviewed literature holds only case reports of them¹³.

What studies found

How often the bone fails to join

A 2020 review gives a figure for waist fractures that are not displaced, or displaced by 0.5 mm or less². After adequate management without surgery, about 90% have united at six weeks². Counts of nonunion differ with the source and the group counted^12,14:

The systematic review of 42 studies: a risk of 2% to 5% in adults¹².
Swiss accident insurance records, 16,691 fractures: about 14% went on to nonunion, and 3% developed arthritis after the injury¹⁴.
Eight UK centres, 250 fractures found on MRI and managed in a cast: 16 (6.3%) went on to delayed union or nonunion¹¹. Delayed union is a broken bone healing more slowly than expected.

What follows a nonunion over the years

Several studies looked at wrists with a nonunion years after the injury^{4,15,16,17,18}.

Who was studied	What was found
55 patients with 56 nonunions and no earlier care for them	Arthritis in 31 of the 32 injured five or more years before¹⁵
102 symptomatic patients with 104 nonunions at least a year old	Osteoarthritis in 22% of nonunions under 5 years old, 75% at 5 to 9 years, and 100% at 10 years or more⁴
33 patients with a nonunion found on X-ray 2 to 37 years after the injury	Progressive osteoarthritis between the radius and the wrist bones in all 33, at re-examination 10 to 17 years later¹⁶
278 consecutive nonunions imaged a mean of 3.3 years after injury	Osteoarthritis in 62.2%¹⁷
20 people with nonunions of 34 years on average and no intervention	Fair or good function in most, despite reduced movement and grip strength¹⁸

In the study of 102 patients, pain did not track the severity of arthritis or the age of the nonunion⁴. Longer duration did go with weaker grip and less wrist motion⁴.

A 1993 appraisal found that none of the natural history studies of scaphoid nonunion published by then followed patients from the time of injury⁵. It concluded that the bias in those studies ran toward a worse outcome⁵. The true course, it said, is less severe than they reported, by an unknown amount⁵.

Surgery against a cast: the SWIFFT trial

SWIFFT was a randomised trial at 31 hospitals in England and Wales³. It was open-label, so participants and researchers knew who was in which group³. The National Institute for Health Research, through its Health Technology Assessment Programme, funded it^3,7. The protocol states that this funder had no part in the study’s design or analysis⁷. The trial assigned 439 adults with waist fractures shifted 2 mm or less to early surgical fixation or a cast³.

The main measure was the patient-rated wrist evaluation (PRWE), a 15-item questionnaire on wrist pain and function^7,22,23. Its total runs from 0, no disability, to 100, and a higher score means worse pain and function^7,22,23. At 52 weeks, the adjusted mean scores were 11.9 with surgery and 14.0 with a cast, not a significant difference³.

In the cast group, a fracture was fixed at once if nonunion was confirmed³. The protocol states that two independent radiologists and a surgeon would judge the X-rays and scans⁷. The same open-label trial, with the same funder, counted complications in each group^3,7.

What the trial counted	Early surgery (219 people)	Cast (220 people)
A potentially serious complication of surgery	31 (14%)³	3 (1%)³
A cast-related complication	5 (2%)³	40 (18%)³
Nonunion at one year	1²⁴	4²⁴

The trial’s report lists a limitation: 17 participants who began in a cast went on to surgery for confirmed nonunion²⁴. Three of the four in the cast group with a nonunion at 52 weeks were not offered surgery²⁴.

At five years, the wrist score still did not differ significantly between the groups²². Among the 267 of 439 who had imaging at five years, osteoarthritis around the scaphoid was similarly common in both groups²³. Three observers read those X-rays and CT scans independently²³. The five-year papers name the same funder, and an independent steering committee oversaw the follow-up^22,23.

Other trials and pooled reviews

Two other randomised comparisons set screw fixation against a cast^25,26. One followed up 71 of its 88 randomised patients at a mean of 93 months²⁵. It found no statistical difference in symptoms and disability, wrist movement, or grip and pinch strength²⁵. The other compared a randomised subgroup with non-displaced waist fractures, 23 in a cast and 15 with a screw²⁶. Time to union did not differ significantly²⁶.

That study assigned groups with a blinded envelope system, and a regional health authority and hospital funded it²⁶. A radiologist and one senior author read all its imaging together, and agreement between readers was not tested²⁶.

Four meta-analyses pooled trials of this one comparison^27,28,29,30. A meta-analysis pools the results of several earlier studies to get one combined answer.

Review	What it found for surgery against a cast
2010, eight randomised trials, 419 patients; fractures not displaced or minimally displaced	Function, measured in four trials, significantly favoured surgery²⁷. No significant difference in nonunion or malunion (a bone that heals out of line)²⁷
2015, six randomised studies in seven papers, 340 fractures not displaced or minimally displaced²⁸	Better short-term function and a quicker return to work after surgery²⁸
2022, seven randomised trials; fractures displaced by 2 mm or less	No significant difference in patient-reported wrist function at 12 months; more complications, and more serious ones, with surgery²⁹
2023, twelve trials, described as randomised	Shorter healing time and return to work after surgery; no significant difference in nonunion or complications³⁰

The 2010 authors wrote that most of the trials lacked scientific rigor²⁷. The 2015 authors wrote that the 2010 review had included one study whose corresponding author confirmed it was not randomised²⁸. The 2015 review counted the 93-month trial above as one study, from its two reports^25,28. For displaced fractures, the authors of a 2012 meta-analysis reported no randomised trials of fixation against a cast at that time³¹.

Return to sport

A 2019 systematic review pooled eleven studies of athletes⁶. Two were randomised trials, six looked back at past records, and three were case series, with no comparison group⁶.

Athletes in the 2019 review	Share who returned to sport	Mean time to return
Managed without surgery	90% (69 of 77)⁶	9.6 weeks⁶
Fixed surgically	98% (81 of 83)⁶	7.3 weeks⁶

The review’s authors noted that the athletes who went straight back in a cast or splint had a nonunion rate of 15%⁶. They wrote that this practice skewed the return times for fractures managed without surgery⁶. They rated their eleven studies’ quality as lower than in similar reviews of other fractures⁶.

A 2026 systematic review found 10 studies covering 208 scaphoid fractures in athletes³². In it, 135 athletes returned on an early timeline, at a mean of 6.25 weeks³². Another 73, managed without surgery, returned after the fracture had united, at a mean of 13.97 weeks³².

Proximal pole fractures were over-represented among those that ended in nonunion, a repeat break or avascular necrosis³². This was most marked in studies that allowed a very early return in a cast³². The authors wrote that the evidence was mostly case series, small cohorts and case reports³². These designs, they wrote, do not allow a reliable estimate of how often complications occur³².

Connections

The bone, the joint and the ligament

The scaphoid meets the end of the radius directly at the radiocarpal joint of the wrist¹. A systematic review of eleven studies looked at injury of the scapholunate ligament alongside an acute scaphoid fracture³³. Reported rates ran from 0% to 71% in studies of low to moderate quality, and complete tears were rare³³.

In an MRI study of 154 people aged 18 to 49 with a wrist sprain and normal X-rays, 13 had a hidden scaphoid fracture³⁴.

Exercise

Related exercise pages: push-up and bench press. A cadaver is a body donated for study. In eight cadaver wrists, an extended-wrist push-up position caused significantly greater peak pressure at the radius–scaphoid part of the joint than a knuckle position³⁵. One case report describes a 21-year-old male cadet with a non-displaced scaphoid waist fracture after he hurt the wrist doing a bench press³⁶.